Network Working Group
Request for Comments: 2760
Category: Informational
M. Allman, Editor
NASA Glenn Research Center/BBN Technologies
S. Dawkins
Nortel
D. Glover
J. Griner
D. Tran
NASA Glenn Research Center
T. Henderson
University of California at Berkeley
J. Heidemann
J. Touch
University of Southern California/ISI
H. Kruse
S. Ostermann
Ohio University
K. Scott
The MITRE Corporation
J. Semke
Pittsburgh Supercomputing Center
February 2000

Ongoing TCP Research Related to Satellites

Status of this Memo

This memo provides information for the Internet community. It does not specify an Internet standard of any kind. Distribution of this memo is unlimited.

Copyright Notice

Copyright © The Internet Society (2000). All Rights Reserved.

Abstract

This document outlines possible TCP enhancements that may allow TCP to better utilize the available bandwidth provided by networks containing satellite links. The algorithms and mechanisms outlined have not been judged to be mature enough to be recommended by the IETF. The goal of this document is to educate researchers as to the current work and progress being done in TCP research related to satellite networks.

Table of Contents

   1         Introduction. . . . . . . . . . . . . . . . . . . .  2
   2         Satellite Architectures . . . . . . . . . . . . . .  3
   2.1       Asymmetric Satellite Networks . . . . . . . . . . .  3
   2.2       Satellite Link as Last Hop. . . . . . . . . . . . .  3
   2.3       Hybrid Satellite Networks     . . . . . . . . . . .  4
   2.4       Point-to-Point Satellite Networks . . . . . . . . .  4
   2.5       Multiple Satellite Hops . . . . . . . . . . . . . .  4
   3         Mitigations . . . . . . . . . . . . . . . . . . . .  4
   3.1       TCP For Transactions. . . . . . . . . . . . . . . .  4
   3.2       Slow Start. . . . . . . . . . . . . . . . . . . . .  5
   3.2.1     Larger Initial Window . . . . . . . . . . . . . . .  6
   3.2.2     Byte Counting . . . . . . . . . . . . . . . . . . .  7
   3.2.3     Delayed ACKs After Slow Start . . . . . . . . . . .  9
   3.2.4     Terminating Slow Start. . . . . . . . . . . . . . . 11
   3.3       Loss Recovery . . . . . . . . . . . . . . . . . . . 12
   3.3.1     Non-SACK Based Mechanisms . . . . . . . . . . . . . 12
   3.3.2     SACK Based Mechanisms . . . . . . . . . . . . . . . 13
   3.3.3     Explicit Congestion Notification. . . . . . . . . . 16
   3.3.4     Detecting Corruption Loss . . . . . . . . . . . . . 18
   3.4       Congestion Avoidance. . . . . . . . . . . . . . . . 21
   3.5       Multiple Data Connections . . . . . . . . . . . . . 22
   3.6       Pacing TCP Segments . . . . . . . . . . . . . . . . 24
   3.7       TCP Header Compression. . . . . . . . . . . . . . . 26
   3.8       Sharing TCP State Among Similar Connections . . . . 29
   3.9       ACK Congestion Control. . . . . . . . . . . . . . . 32
   3.10      ACK Filtering . . . . . . . . . . . . . . . . . . . 34
   4         Conclusions . . . . . . . . . . . . . . . . . . . . 36
   5         Security Considerations . . . . . . . . . . . . . . 36
   6         Acknowledgments . . . . . . . . . . . . . . . . . . 37
   7         References. . . . . . . . . . . . . . . . . . . . . 37
   8         Authors' Addresses. . . . . . . . . . . . . . . . . 43
   9         Full Copyright Statement. . . . . . . . . . . . . . 46

1 Introduction

This document outlines mechanisms that may help the Transmission Control Protocol (TCP) [Pos81] better utilize the bandwidth provided by long-delay satellite environments. These mechanisms may also help in other environments or for other protocols. The proposals outlined in this document are currently being studied throughout the research community. Therefore, these mechanisms are not mature enough to be recommended for wide-spread use by the IETF. However, some of these mechanisms may be safely used today. It is hoped that this document will stimulate further study into the described mechanisms. If, at some point, the mechanisms discussed in this memo prove to be safe and appropriate to be recommended for general use, the appropriate IETF documents will be written.

It should be noted that non-TCP mechanisms that help performance over satellite links do exist (e.g., application-level changes, queueing disciplines, etc.). However, outlining these non-TCP mitigations is beyond the scope of this document and therefore is left as future work. Additionally, there are a number of mitigations to TCP's performance problems that involve very active intervention by gateways along the end-to-end path from the sender to the receiver. Documenting the pros and cons of such solutions is also left as future work.

2 Satellite Architectures

Specific characteristics of satellite links and the impact these characteristics have on TCP are presented in RFC 2488 [AGS99]. This section discusses several possible topologies where satellite links may be integrated into the global Internet. The mitigation outlined in section 3 will include a discussion of which environment the mechanism is expected to benefit.

2.1 Asymmetric Satellite Networks

Some satellite networks exhibit a bandwidth asymmetry, a larger data rate in one direction than the reverse direction, because of limits on the transmission power and the antenna size at one end of the
link. Meanwhile, some other satellite systems are unidirectional and use a non-satellite return path (such as a dialup modem link). The nature of most TCP traffic is asymmetric with data flowing in one direction and acknowledgments in opposite direction. However, the term asymmetric in this document refers to different physical capacities in the forward and return links. Asymmetry has been shown to be a problem for TCP [BPK97,BPK98].

2.2 Satellite Link as Last Hop

Satellite links that provide service directly to end users, as opposed to satellite links located in the middle of a network, may allow for specialized design of protocols used over the last hop. Some satellite providers use the satellite link as a shared high speed downlink to users with a lower speed, non-shared terrestrial link that is used as a return link for requests and acknowledgments. Many times this creates an asymmetric network, as discussed above.

2.3 Hybrid Satellite Networks

In the more general case, satellite links may be located at any point in the network topology. In this case, the satellite link acts as just another link between two gateways. In this environment, a given connection may be sent over terrestrial links (including terrestrial wireless), as well as satellite links. On the other hand, a connection could also travel over only the terrestrial network or only over the satellite portion of the network.

2.4 Point-to-Point Satellite Networks

In point-to-point satellite networks, the only hop in the network is over the satellite link. This pure satellite environment exhibits only the problems associated with the satellite links, as outlined in [AGS99]. Since this is a private network, some mitigations that are not appropriate for shared networks can be considered.

2.5 Multiple Satellite Hops

In some situations, network traffic may traverse multiple satellite hops between the source and the destination. Such an environment aggravates the satellite characteristics described in [AGS99].

3 Mitigations

The following sections will discuss various techniques for mitigating the problems TCP faces in the satellite environment. Each of the following sections will be organized as follows: First, each mitigation will be briefly outlined. Next, research work involving the mechanism in question will be briefly discussed. Next the implementation issues of the mechanism will be presented (including whether or not the particular mechanism presents any dangers to shared networks). Then a discussion of the mechanism's potential with regard to the topologies outlined above is given. Finally, the relationships and possible interactions with other TCP mechanisms are outlined. The reader is expected to be familiar with the TCP terminology used in [AGS99].

3.1 TCP For Transactions

3.1.1 Mitigation Description

TCP uses a three-way handshake to setup a connection between two hosts [Pos81]. This connection setup requires 1-1.5 round-trip times (RTTs), depending upon whether the data sender started the connection actively or passively. This startup time can be eliminated by using TCP extensions for transactions (T/TCP) [Bra94]. After the first connection between a pair of hosts is established, T/TCP is able to bypass the three-way handshake, allowing the data sender to begin transmitting data in the first segment sent (along with the SYN). This is especially helpful for short request/response traffic, as it saves a potentially long setup phase when no useful data is being transmitted.

3.1.2 Research

T/TCP is outlined and analyzed in [Bra92,Bra94].

3.1.3 Implementation Issues

T/TCP requires changes in the TCP stacks of both the data sender and the data receiver. While T/TCP is safe to implement in shared networks from a congestion control perspective, several security implications of sending data in the first data segment have been identified [ddKI99].

3.1.4 Topology Considerations

It is expected that T/TCP will be equally beneficial in all environments outlined in section 2.

3.1.5 Possible Interaction and Relationships with Other Research

T/TCP allows data transfer to start more rapidly, much like using a larger initial congestion window (see section 3.2.1), delayed ACKs after slow start (section 3.2.3) or byte counting (section 3.2.2).

3.2 Slow Start

The slow start algorithm is used to gradually increase the size of TCP's congestion window (cwnd) [Jac88,Ste97,APS99]. The algorithm is an important safe-guard against transmitting an inappropriate amount of data into the network when the connection starts up. However, slow start can also waste available network capacity, especially in long-delay networks [All97a,Hay97]. Slow start is particularly inefficient for transfers that are short compared to the delay*bandwidth product of the network (e.g., WWW transfers).

Delayed ACKs are another source of wasted capacity during the slow start phase. RFC 1122 [Bra89] suggests data receivers refrain from ACKing every incoming data segment. However, every second full-sized segment should be ACKed. If a second full-sized segment does not arrive within a given timeout, an ACK must be generated (this timeout cannot exceed 500 ms). Since the data sender increases the size of cwnd based on the number of arriving ACKs, reducing the number of

ACKs slows the cwnd growth rate. In addition, when TCP starts sending, it sends 1 segment. When using delayed ACKs a second segment must arrive before an ACK is sent. Therefore, the receiver is always forced to wait for the delayed ACK timer to expire before ACKing the first segment, which also increases the transfer time.

Several proposals have suggested ways to make slow start less time consuming. These proposals are briefly outlined below and references to the research work given.

3.2.1 Larger Initial Window

3.2.1.1 Mitigation Description

One method that will reduce the amount of time required by slow start (and therefore, the amount of wasted capacity) is to increase the initial value of cwnd. An experimental TCP extension outlined in [AFP98] allows the initial size of cwnd to be increased from 1 segment to that given in equation (1).

               min (4*MSS, max (2*MSS, 4380 bytes))               (1)

By increasing the initial value of cwnd, more packets are sent during the first RTT of data transmission, which will trigger more ACKs, allowing the congestion window to open more rapidly. In addition, by sending at least 2 segments initially, the first segment does not need to wait for the delayed ACK timer to expire as is the case when the initial size of cwnd is 1 segment (as discussed above). Therefore, the value of cwnd given in equation 1 saves up to 3 RTTs and a delayed ACK timeout when compared to an initial cwnd of 1 segment.

Also, we note that RFC 2581 [APS99], a standards-track document, allows a TCP to use an initial cwnd of up to 2 segments. This change is highly recommended for satellite networks.

3.2.1.2 Research

Several researchers have studied the use of a larger initial window in various environments. [Nic97] and [KAGT98] show a reduction in WWW page transfer time over hybrid fiber coax (HFC) and satellite links respectively. Furthermore, it has been shown that using an initial cwnd of 4 segments does not negatively impact overall performance over dialup modem links with a small number of buffers [SP98]. [AHO98] shows an improvement in transfer time for 16 KB files across the Internet and dialup modem links when using a larger initial value for cwnd. However, a slight increase in dropped segments was also shown. Finally, [PN98] shows improved transfer time for WWW traffic in simulations with competing traffic, in addition to a small increase in the drop rate.

3.2.1.3 Implementation Issues

The use of a larger initial cwnd value requires changes to the sender's TCP stack. Using an initial congestion window of 2 segments is allowed by RFC 2581 [APS99]. Using an initial congestion window of 3 or 4 segments is not expected to present any danger of congestion collapse [AFP98], however may degrade performance in some networks.

3.2.1.4 Topology Considerations

It is expected that the use of a large initial window would be equally beneficial to all network architectures outlined in section
2.

3.2.1.5 Possible Interaction and Relationships with Other Research

Using a fixed larger initial congestion window decreases the impact of a long RTT on transfer time (especially for short transfers) at the cost of bursting data into a network with unknown conditions. A mechanism that mitigates bursts may make the use of a larger initial congestion window more appropriate (e.g., limiting the size of line- rate bursts [FF96] or pacing the segments in a burst [VH97a]).

Also, using delayed ACKs only after slow start (as outlined in section 3.2.3) offers an alternative way to immediately ACK the first segment of a transfer and open the congestion window more rapidly. Finally, using some form of TCP state sharing among a number of connections (as discussed in 3.8) may provide an alternative to using a fixed larger initial window.

3.2.2 Byte Counting

3.2.2.1 Mitigation Description

As discussed above, the wide-spread use of delayed ACKs increases the time needed by a TCP sender to increase the size of the congestion window during slow start. This is especially harmful to flows traversing long-delay GEO satellite links. One mechanism that has been suggested to mitigate the problems caused by delayed ACKs is the use of "byte counting", rather than standard ACK counting [All97a,All98]. Using standard ACK counting, the congestion window is increased by 1 segment for each ACK received during slow start. However, using byte counting the congestion window increase is based on the number of previously unacknowledged bytes covered by each incoming ACK, rather than on the number of ACKs received. This makes the increase relative to the amount of data transmitted, rather than being dependent on the ACK interval used by the receiver.

Two forms of byte counting are studied in [All98]. The first is unlimited byte counting (UBC). This mechanism simply uses the number of previously unacknowledged bytes to increase the congestion window each time an ACK arrives. The second form is limited byte counting (LBC). LBC limits the amount of cwnd increase to 2 segments. This limit throttles the size of the burst of data sent in response to a "stretch ACK" [Pax97]. Stretch ACKs are acknowledgments that cover more than 2 segments of previously unacknowledged data. Stretch ACKs can occur by design [Joh95] (although this is not standard), due to implementation bugs [All97b,PADHV99] or due to ACK loss. [All98] shows that LBC prevents large line-rate bursts when compared to UBC, and therefore offers fewer dropped segments and better performance. In addition, UBC causes large bursts during slow start based loss recovery due to the large cumulative ACKs that can arrive during loss recovery. The behavior of UBC during loss recovery can cause large decreases in performance and [All98] strongly recommends UBC not be deployed without further study into mitigating the large bursts.

Note: The standards track RFC 2581 [APS99] allows a TCP to use byte counting to increase cwnd during congestion avoidance, however not during slow start.

3.2.2.2 Research

Using byte counting, as opposed to standard ACK counting, has been shown to reduce the amount of time needed to increase the value of cwnd to an appropriate size in satellite networks [All97a]. In addition, [All98] presents a simulation comparison of byte counting and the standard cwnd increase algorithm in uncongested networks and networks with competing traffic. This study found that the limited form of byte counting outlined above can improve performance, while also increasing the drop rate slightly.

[BPK97,BPK98] also investigated unlimited byte counting in conjunction with various ACK filtering algorithms (discussed in section 3.10) in asymmetric networks.

3.2.2.3 Implementation Issues

Changing from ACK counting to byte counting requires changes to the data sender's TCP stack. Byte counting violates the algorithm for increasing the congestion window outlined in RFC 2581 [APS99] (by making congestion window growth more aggressive during slow start) and therefore should not be used in shared networks.

3.2.2.4 Topology Considerations

It has been suggested by some (and roundly criticized by others) that byte counting will allow TCP to provide uniform cwnd increase, regardless of the ACKing behavior of the receiver. In addition, byte counting also mitigates the retarded window growth provided by receivers that generate stretch ACKs because of the capacity of the return link, as discussed in [BPK97,BPK98]. Therefore, this change is expected to be especially beneficial to asymmetric networks.

3.2.2.5 Possible Interaction and Relationships with Other Research

Unlimited byte counting should not be used without a method to mitigate the potentially large line-rate bursts the algorithm can cause. Also, LBC may send bursts that are too large for the given network conditions. In this case, LBC may also benefit from some algorithm that would lessen the impact of line-rate bursts of segments. Also note that using delayed ACKs only after slow start (as outlined in section 3.2.3) negates the limited byte counting algorithm because each ACK covers only one segment during slow start. Therefore, both ACK counting and byte counting yield the same increase in the congestion window at this point (in the first RTT).

3.2.3 Delayed ACKs After Slow Start

3.2.3.1 Mitigation Description

As discussed above, TCP senders use the number of incoming ACKs to increase the congestion window during slow start. And, since delayed ACKs reduce the number of ACKs returned by the receiver by roughly half, the rate of growth of the congestion window is reduced. One proposed solution to this problem is to use delayed ACKs only after the slow start (DAASS) phase. This provides more ACKs while TCP is aggressively increasing the congestion window and less ACKs while TCP is in steady state, which conserves network resources.

3.2.3.2 Research

[All98] shows that in simulation, using delayed ACKs after slow start (DAASS) improves transfer time when compared to a receiver that always generates delayed ACKs. However, DAASS also slightly increases the loss rate due to the increased rate of cwnd growth.

3.2.3.3 Implementation Issues

The major problem with DAASS is in the implementation. The receiver has to somehow know when the sender is using the slow start
algorithm. The receiver could implement a heuristic that attempts to watch the change in the amount of data being received and change the ACKing behavior accordingly. Or, the sender could send a message (a flipped bit in the TCP header, perhaps) indicating that it was using slow start. The implementation of DAASS is, therefore, an open
issue.

Using DAASS does not violate the TCP congestion control specification [APS99]. However, the standards (RFC 2581 [APS99]) currently recommend using delayed acknowledgments and DAASS goes (partially) against this recommendation.

3.2.3.4 Topology Considerations

DAASS should work equally well in all scenarios presented in section 2. However, in asymmetric networks it may aggravate ACK congestion in the return link, due to the increased number of ACKs (see sections 3.9 and 3.10 for a more detailed discussion of ACK congestion).

3.2.3.5 Possible Interaction and Relationships with Other Research

DAASS has several possible interactions with other proposals made in the research community. DAASS can aggravate congestion on the path between the data receiver and the data sender due to the increased number of returning acknowledgments. This can have an especially adverse effect on asymmetric networks that are prone to experiencing ACK congestion. As outlined in sections 3.9 and 3.10, several mitigations have been proposed to reduce the number of ACKs that are passed over a low-bandwidth return link. Using DAASS will increase the number of ACKs sent by the receiver. The interaction between DAASS and the methods for reducing the number of ACKs is an open research question. Also, as noted in section 3.2.1.5 above, DAASS provides some of the same benefits as using a larger initial congestion window and therefore it may not be desirable to use both mechanisms together. However, this remains an open question. Finally, DAASS and limited byte counting are both used to increase the rate at which the congestion window is opened. The DAASS algorithm substantially reduces the impact limited byte counting has on the rate of congestion window increase.

3.2.4 Terminating Slow Start

3.2.4.1 Mitigation Description

The initial slow start phase is used by TCP to determine an appropriate congestion window size for the given network conditions [Jac88]. Slow start is terminated when TCP detects congestion, or when the size of cwnd reaches the size of the receiver's advertised window. Slow start is also terminated if cwnd grows beyond a certain size. The threshold at which TCP ends slow start and begins using the congestion avoidance algorithm is called "ssthresh" [Jac88]. In most implementations, the initial value for ssthresh is the receiver's advertised window. During slow start, TCP roughly doubles the size of cwnd every RTT and therefore can overwhelm the network with at most twice as many segments as the network can handle. By setting ssthresh to a value less than the receiver's advertised window initially, the sender may avoid overwhelming the network with twice the appropriate number of segments. Hoe [Hoe96] proposes using the packet-pair algorithm [Kes91] and the measured RTT to determine a more appropriate value for ssthresh. The algorithm observes the spacing between the first few returning ACKs to determine the bandwidth of the bottleneck link. Together with the measured RTT, the delay*bandwidth product is determined and ssthresh is set to this value. When TCP's cwnd reaches this reduced ssthresh, slow start is terminated and transmission continues using congestion avoidance, which is a more conservative algorithm for increasing the size of the congestion window.

3.2.4.2 Research

It has been shown that estimating ssthresh can improve performance and decrease packet loss in simulations [Hoe96]. However, obtaining an accurate estimate of the available bandwidth in a dynamic network is very challenging, especially attempting to do so on the sending side of the TCP connection [AP99]. Therefore, before this mechanism is widely deployed, bandwidth estimation must be studied in a more detail.

3.2.4.3 Implementation Issues

As outlined in [Hoe96], estimating ssthresh requires changes to the data sender's TCP stack. As suggested in [AP99], bandwidth estimates may be more accurate when taken by the TCP receiver, and therefore both sender and receiver changes would be required. Estimating ssthresh is safe to implement in production networks from a congestion control perspective, as it can only make TCP more conservative than outlined in RFC 2581 [APS99] (assuming the TCP implementation is using an initial ssthresh of infinity as allowed by [APS99]).

3.2.4.4 Topology Considerations

It is expected that this mechanism will work equally well in all symmetric topologies outlined in section 2. However, asymmetric links pose a special problem, as the rate of the returning ACKs may not be the bottleneck bandwidth in the forward direction. This can lead to the sender setting ssthresh too low. Premature termination of slow start can hurt performance, as congestion avoidance opens cwnd more conservatively. Receiver-based bandwidth estimators do not suffer from this problem.

3.2.4.5 Possible Interaction and Relationships with Other Research

Terminating slow start at the right time is useful to avoid multiple dropped segments. However, using a selective acknowledgment-based loss recovery scheme (as outlined in section 3.3.2) can drastically improve TCP's ability to quickly recover from multiple lost segments Therefore, it may not be as important to terminate slow start before a large loss event occurs. [AP99] shows that using delayed acknowledgments [Bra89] reduces the effectiveness of sender-side bandwidth estimation. Therefore, using delayed ACKs only during slow start (as outlined in section 3.2.3) may make bandwidth estimation more feasible.

3.3 Loss Recovery

3.3.1 Non-SACK Based Mechanisms

3.3.1.1 Mitigation Description

Several similar algorithms have been developed and studied that improve TCP's ability to recover from multiple lost segments in a window of data without relying on the (often long) retransmission timeout. These sender-side algorithms, known as NewReno TCP, do not depend on the availability of selective acknowledgments (SACKs) [MMFR96].

These algorithms generally work by updating the fast recovery algorithm to use information provided by "partial ACKs" to trigger retransmissions. A partial ACK covers some new data, but not all data outstanding when a particular loss event starts. For instance, consider the case when segment N is retransmitted using the fast retransmit algorithm and segment M is the last segment sent when segment N is resent. If segment N is the only segment lost, the ACK elicited by the retransmission of segment N would be for segment M. If, however, segment N+1 was also lost, the ACK elicited by the retransmission of segment N will be N+1. This can be taken as an indication that segment N+1 was lost and used to trigger a retransmission.

3.3.1.2 Research

Hoe [Hoe95,Hoe96] introduced the idea of using partial ACKs to trigger retransmissions and showed that doing so could improve performance. [FF96] shows that in some cases using partial ACKs to trigger retransmissions reduces the time required to recover from multiple lost segments. However, [FF96] also shows that in some cases (many lost segments) relying on the RTO timer can improve performance over simply using partial ACKs to trigger all retransmissions. [HK99] shows that using partial ACKs to trigger retransmissions, in conjunction with SACK, improves performance when compared to TCP using fast retransmit/fast recovery in a satellite environment. Finally, [FH99] describes several slightly different variants of NewReno.

3.3.1.3 Implementation Issues

Implementing these fast recovery enhancements requires changes to the sender-side TCP stack. These changes can safely be implemented in production networks and are allowed by RFC 2581 [APS99].

3.3.1.4 Topology Considerations

It is expected that these changes will work well in all environments outlined in section 2.

3.3.1.5 Possible Interaction and Relationships with Other Research

See section 3.3.2.2.5.

3.3.2 SACK Based Mechanisms

3.3.2.1 Fast Recovery with SACK
3.3.2.1.1 Mitigation Description

Fall and Floyd [FF96] describe a conservative extension to the fast recovery algorithm that takes into account information provided by selective acknowledgments (SACKs) [MMFR96] sent by the receiver. The algorithm starts after fast retransmit triggers the resending of a segment. As with fast retransmit, the algorithm cuts cwnd in half when a loss is detected. The algorithm keeps a variable called "pipe", which is an estimate of the number of outstanding segments in the network. The pipe variable is decremented by 1 segment for each duplicate ACK that arrives with new SACK information. The pipe variable is incremented by 1 for each new or retransmitted segment sent. A segment may be sent when the value of pipe is less than cwnd (this segment is either a retransmission per the SACK information or a new segment if the SACK information indicates that no more retransmits are needed).

This algorithm generally allows TCP to recover from multiple segment losses in a window of data within one RTT of loss detection. Like the forward acknowledgment (FACK) algorithm described below, the SACK information allows the pipe algorithm to decouple the choice of when to send a segment from the choice of what segment to send.

[APS99] allows the use of this algorithm, as it is consistent with the spirit of the fast recovery algorithm.

3.3.2.1.2 Research

[FF96] shows that the above described SACK algorithm performs better than several non-SACK based recovery algorithms when 1--4 segments are lost from a window of data. [AHKO97] shows that the algorithm improves performance over satellite links. Hayes [Hay97] shows the in certain circumstances, the SACK algorithm can hurt performance by generating a large line-rate burst of data at the end of loss recovery, which causes further loss.

3.3.2.1.3 Implementation Issues

This algorithm is implemented in the sender's TCP stack. However, it relies on SACK information generated by the receiver. This algorithm is safe for shared networks and is allowed by RFC 2581 [APS99].

3.3.2.1.4 Topology Considerations

It is expected that the pipe algorithm will work equally well in all scenarios presented in section 2.

3.3.2.1.5 Possible Interaction and Relationships with Other Research

See section 3.3.2.2.5.

3.3.2.2 Forward Acknowledgments
3.3.2.2.1 Mitigation Description

The Forward Acknowledgment (FACK) algorithm [MM96a,MM96b] was developed to improve TCP congestion control during loss recovery. FACK uses TCP SACK options to glean additional information about the congestion state, adding more precise control to the injection of data into the network during recovery. FACK decouples the congestion control algorithms from the data recovery algorithms to provide a simple and direct way to use SACK to improve congestion control. Due to the separation of these two algorithms, new data may be sent during recovery to sustain TCP's self-clock when there is no further data to retransmit.

The most recent version of FACK is Rate-Halving [MM96b], in which one packet is sent for every two ACKs received during recovery. Transmitting a segment for every-other ACK has the result of reducing the congestion window in one round trip to half of the number of packets that were successfully handled by the network (so when cwnd is too large by more than a factor of two it still gets reduced to half of what the network can sustain). Another important aspect of FACK with Rate-Halving is that it sustains the ACK self-clock during recovery because transmitting a packet for every-other ACK does not require half a cwnd of data to drain from the network before transmitting, as required by the fast recovery algorithm
[Ste97,APS99].

In addition, the FACK with Rate-Halving implementation provides Thresholded Retransmission to each lost segment. "Tcprexmtthresh" is the number of duplicate ACKs required by TCP to trigger a fast retransmit and enter recovery. FACK applies thresholded retransmission to all segments by waiting until tcprexmtthresh SACK blocks indicate that a given segment is missing before resending the segment. This allows reasonable behavior on links that reorder segments. As described above, FACK sends a segment for every second ACK received during recovery. New segments are transmitted except when tcprexmtthresh SACK blocks have been observed for a dropped segment, at which point the dropped segment is retransmitted.

[APS99] allows the use of this algorithm, as it is consistent with the spirit of the fast recovery algorithm.

3.3.2.2.2 Research

The original FACK algorithm is outlined in [MM96a]. The algorithm was later enhanced to include Rate-Halving [MM96b]. The real-world performance of FACK with Rate-Halving was shown to be much closer to the theoretical maximum for TCP than either TCP Reno or the SACK- based extensions to fast recovery outlined in section 3.3.2.1
[MSMO97].

3.3.2.2.3 Implementation Issues

In order to use FACK, the sender's TCP stack must be modified. In addition, the receiver must be able to generate SACK options to obtain the full benefit of using FACK. The FACK algorithm is safe for shared networks and is allowed by RFC 2581 [APS99].

3.3.2.2.4 Topology Considerations

FACK is expected to improve performance in all environments outlined in section 2. Since it is better able to sustain its self-clock than TCP Reno, it may be considerably more attractive over long delay
paths.

3.3.2.2.5 Possible Interaction and Relationships with Other Research

Both SACK based loss recovery algorithms described above (the fast recovery enhancement and the FACK algorithm) are similar in that they attempt to effectively repair multiple lost segments from a window of data. Which of the SACK-based loss recovery algorithms to use is still an open research question. In addition, these algorithms are similar to the non-SACK NewReno algorithm described in section 3.3.1, in that they attempt to recover from multiple lost segments without reverting to using the retransmission timer. As has been shown, the above SACK based algorithms are more robust than the NewReno algorithm. However, the SACK algorithm requires a cooperating TCP receiver, which the NewReno algorithm does not. A reasonable TCP implementation might include both a SACK-based and a NewReno-based loss recovery algorithm such that the sender can use the most appropriate loss recovery algorithm based on whether or not the receiver supports SACKs. Finally, both SACK-based and non-SACK-based versions of fast recovery have been shown to transmit a large burst of data upon leaving loss recovery, in some cases [Hay97]. Therefore, the algorithms may benefit from some burst suppression algorithm.

3.3.3 Explicit Congestion Notification

3.3.3.1 Mitigation Description

Explicit congestion notification (ECN) allows routers to inform TCP senders about imminent congestion without dropping segments. Two major forms of ECN have been studied. A router employing backward ECN (BECN), transmits messages directly to the data originator informing it of congestion. IP routers can accomplish this with an ICMP Source Quench message. The arrival of a BECN signal may or may not mean that a TCP data segment has been dropped, but it is a clear indication that the TCP sender should reduce its sending rate (i.e., the value of cwnd). The second major form of congestion notification is forward ECN (FECN). FECN routers mark data segments with a special tag when congestion is imminent, but forward the data
segment. The data receiver then echos the congestion information back to the sender in the ACK packet. A description of a FECN mechanism for TCP/IP is given in [RF99].

As described in [RF99], senders transmit segments with an "ECN- Capable Transport" bit set in the IP header of each packet. If a router employing an active queueing strategy, such as Random Early Detection (RED) [FJ93,BCC+98], would otherwise drop this segment, an "Congestion Experienced" bit in the IP header is set instead. Upon reception, the information is echoed back to TCP senders using a bit in the TCP header. The TCP sender adjusts the congestion window just as it would if a segment was dropped.

The implementation of ECN as specified in [RF99] requires the deployment of active queue management mechanisms in the affected routers. This allows the routers to signal congestion by sending TCP a small number of "congestion signals" (segment drops or ECN messages), rather than discarding a large number of segments, as can happen when TCP overwhelms a drop-tail router queue.

Since satellite networks generally have higher bit-error rates than terrestrial networks, determining whether a segment was lost due to congestion or corruption may allow TCP to achieve better performance in high BER environments than currently possible (due to TCP's assumption that all loss is due to congestion). While not a solution to this problem, adding an ECN mechanism to TCP may be a part of a mechanism that will help achieve this goal. See section 3.3.4 for a more detailed discussion of differentiating between corruption and congestion based losses.

3.3.3.2 Research

[Flo94] shows that ECN is effective in reducing the segment loss rate which yields better performance especially for short and interactive TCP connections. Furthermore, [Flo94] also shows that ECN avoids some unnecessary, and costly TCP retransmission timeouts. Finally, [Flo94] also considers some of the advantages and disadvantages of various forms of explicit congestion notification.

3.3.3.3 Implementation Issues

Deployment of ECN requires changes to the TCP implementation on both sender and receiver. Additionally, deployment of ECN requires deployment of some active queue management infrastructure in routers. RED is assumed in most ECN discussions, because RED is already identifying segments to drop, even before its buffer space is exhausted. ECN simply allows the delivery of "marked" segments while still notifying the end nodes that congestion is occurring along the path. ECN is safe (from a congestion control perspective) for shared networks, as it maintains the same TCP congestion control principles as are used when congestion is detected via segment drops.

3.3.3.4 Topology Considerations

It is expected that none of the environments outlined in section 2 will present a bias towards or against ECN traffic.

3.3.3.5 Possible Interaction and Relationships with Other Research

Note that some form of active queueing is necessary to use ECN (e.g., RED queueing).

3.3.4 Detecting Corruption Loss

Differentiating between congestion (loss of segments due to router buffer overflow or imminent buffer overflow) and corruption (loss of segments due to damaged bits) is a difficult problem for TCP. This differentiation is particularly important because the action that TCP should take in the two cases is entirely different. In the case of corruption, TCP should merely retransmit the damaged segment as soon as its loss is detected; there is no need for TCP to adjust its congestion window. On the other hand, as has been widely discussed above, when the TCP sender detects congestion, it should immediately reduce its congestion window to avoid making the congestion worse.

TCP's defined behavior, as motivated by [Jac88,Jac90] and defined in [Bra89,Ste97,APS99], is to assume that all loss is due to congestion and to trigger the congestion control algorithms, as defined in [Ste97,APS99]. The loss may be detected using the fast retransmit algorithm, or in the worst case is detected by the expiration of TCP's retransmission timer.

TCP's assumption that loss is due to congestion rather than corruption is a conservative mechanism that prevents congestion collapse [Jac88,FF98]. Over satellite networks, however, as in many wireless environments, loss due to corruption is more common than on terrestrial networks. One common partial solution to this problem is to add Forward Error Correction (FEC) to the data that's sent over the satellite/wireless link. A more complete discussion of the benefits of FEC can be found in [AGS99]. However, given that FEC does not always work or cannot be universally applied, other mechanisms have been studied to attempt to make TCP able to differentiate between congestion-based and corruption-based loss.

TCP segments that have been corrupted are most often dropped by intervening routers when link-level checksum mechanisms detect that an incoming frame has errors. Occasionally, a TCP segment containing an error may survive without detection until it arrives at the TCP receiving host, at which point it will almost always either fail the IP header checksum or the TCP checksum and be discarded as in the link-level error case. Unfortunately, in either of these cases, it's not generally safe for the node detecting the corruption to return information about the corrupt packet to the TCP sender because the sending address itself might have been corrupted.

3.3.4.1 Mitigation Description

Because the probability of link errors on a satellite link is relatively greater than on a hardwired link, it is particularly important that the TCP sender retransmit these lost segments without reducing its congestion window. Because corrupt segments do not indicate congestion, there is no need for the TCP sender to enter a congestion avoidance phase, which may waste available bandwidth. Simulations performed in [SF98] show a performance improvement when TCP can properly differentiate between between corruption and congestion of wireless links.

Perhaps the greatest research challenge in detecting corruption is getting TCP (a transport-layer protocol) to receive appropriate information from either the network layer (IP) or the link layer. Much of the work done to date has involved link-layer mechanisms that retransmit damaged segments. The challenge seems to be to get these mechanisms to make repairs in such a way that TCP understands what happened and can respond appropriately.

3.3.4.2 Research

Research into corruption detection to date has focused primarily on making the link level detect errors and then perform link-level retransmissions. This work is summarized in [BKVP97,BPSK96]. One of the problems with this promising technique is that it causes an effective reordering of the segments from the TCP receiver's point of view. As a simple example, if segments A B C D are sent across a noisy link and segment B is corrupted, segments C and D may have already crossed the link before B can be retransmitted at the link level, causing them to arrive at the TCP receiver in the order A C D B. This segment reordering would cause the TCP receiver to generate duplicate ACKs upon the arrival of segments C and D. If the reordering was bad enough, the sender would trigger the fast retransmit algorithm in the TCP sender, in response to the duplicate ACKs. Research presented in [MV98] proposes the idea of suppressing or delaying the duplicate ACKs in the reverse direction to counteract this behavior. Alternatively, proposals that make TCP more robust in the face of re-ordered segment arrivals [Flo99] may reduce the side effects of the re-ordering caused by link-layer retransmissions.

A more high-level approach, outlined in the [DMT96], uses a new "corruption experienced" ICMP error message generated by routers that detect corruption. These messages are sent in the forward direction, toward the packet's destination, rather than in the reverse direction as is done with ICMP Source Quench messages. Sending the error messages in the forward direction allows this feedback to work over asymmetric paths. As noted above, generating an error message in response to a damaged packet is problematic because the source and destination addresses may not be valid. The mechanism outlined in [DMT96] gets around this problem by having the routers maintain a small cache of recent packet destinations; when the router experiences an error rate above some threshold, it sends an ICMP corruption-experienced message to all of the destinations in its cache. Each TCP receiver then must return this information to its respective TCP sender (through a TCP option). Upon receiving an ACK with this "corruption-experienced" option, the TCP sender assumes that packet loss is due to corruption rather than congestion for two round trip times (RTT) or until it receives additional link state information (such as "link down", source quench, or additional "corruption experienced" messages). Note that in shared networks, ignoring segment loss for 2 RTTs may aggravate congestion by making TCP unresponsive.

3.3.4.3 Implementation Issues

All of the techniques discussed above require changes to at least the TCP sending and receiving stacks, as well as intermediate routers. Due to the concerns over possibly ignoring congestion signals (i.e., segment drops), the above algorithm is not recommended for use in shared networks.

3.3.4.4 Topology Considerations

It is expected that corruption detection, in general would be beneficial in all environments outlined in section 2. It would be particularly beneficial in the satellite/wireless environment over which these errors may be more prevalent.

3.3.4.5 Possible Interaction and Relationships with Other Research

SACK-based loss recovery algorithms (as described in 3.3.2) may reduce the impact of corrupted segments on mostly clean links because recovery will be able to happen more rapidly (and without relying on the retransmission timer). Note that while SACK-based loss recovery helps, throughput will still suffer in the face of non-congestion related packet loss.

3.4 Congestion Avoidance

3.4.1 Mitigation Description

During congestion avoidance, in the absence of loss, the TCP sender adds approximately one segment to its congestion window during each RTT [Jac88,Ste97,APS99]. Several researchers have observed that this policy leads to unfair sharing of bandwidth when multiple connections with different RTTs traverse the same bottleneck link, with the long RTT connections obtaining only a small fraction of their fair share of the bandwidth.

One effective solution to this problem is to deploy fair queueing and TCP-friendly buffer management in network routers [Sut98]. However, in the absence of help from the network, other researchers have investigated changes to the congestion avoidance policy at the TCP sender, as described in [Flo91,HK98].

3.4.2 Research

The "Constant-Rate" increase policy has been studied in [Flo91,HK98]. It attempts to equalize the rate at which TCP senders increase their sending rate during congestion avoidance. Both [Flo91] and [HK98] illustrate cases in which the "Constant-Rate" policy largely corrects the bias against long RTT connections, although [HK98] presents some evidence that such a policy may be difficult to incrementally deploy in an operational network. The proper selection of a constant (for the constant rate of increase) is an open issue.

The "Increase-by-K" policy can be selectively used by long RTT connections in a heterogeneous environment. This policy simply changes the slope of the linear increase, with connections over a given RTT threshold adding "K" segments to the congestion window every RTT, instead of one. [HK98] presents evidence that this policy, when used with small values of "K", may be successful in reducing the unfairness while keeping the link utilization high, when a small number of connections share a bottleneck link. The selection of the constant "K," the RTT threshold to invoke this policy, and performance under a large number of flows are all open issues.

3.4.3 Implementation Issues

Implementation of either the "Constant-Rate" or "Increase-by-K" policies requires a change to the congestion avoidance mechanism at the TCP sender. In the case of "Constant-Rate," such a change must be implemented globally. Additionally, the TCP sender must have a reasonably accurate estimate of the RTT of the connection. The algorithms outlined above violate the congestion avoidance algorithm as outlined in RFC 2581 [APS99] and therefore should not be implemented in shared networks at this time.

3.4.4 Topology Considerations

These solutions are applicable to all satellite networks that are integrated with a terrestrial network, in which satellite connections may be competing with terrestrial connections for the same bottleneck link.

3.4.5 Possible Interaction and Relationships with Other Research

As shown in [PADHV99], increasing the congestion window by multiple segments per RTT can cause TCP to drop multiple segments and force a retransmission timeout in some versions of TCP. Therefore, the above changes to the congestion avoidance algorithm may need to be accompanied by a SACK-based loss recovery algorithm that can quickly repair multiple dropped segments.

3.5 Multiple Data Connections

3.5.1 Mitigation Description

One method that has been used to overcome TCP's inefficiencies in the satellite environment is to use multiple TCP flows to transfer a given file. The use of N TCP connections makes the sender N times more aggressive and therefore can improve throughput in some situations. Using N multiple TCP connections can impact the transfer and the network in a number of ways, which are listed below.

  1. The transfer is able to start transmission using an effective congestion window of N segments, rather than a single segment as one TCP flow uses. This allows the transfer to more quickly increase the effective cwnd size to an appropriate size for the given network. However, in some circumstances an initial window of N segments is inappropriate for the network conditions. In this case, a transfer utilizing more than one connection may aggravate congestion.
  1. During the congestion avoidance phase, the transfer increases the effective cwnd by N segments per RTT, rather than the one segment per RTT increase that a single TCP connection provides. Again, this can aid the transfer by more rapidly increasing the effective cwnd to an appropriate point. However, this rate of increase can also be too aggressive for the network conditions. In this case, the use of multiple data connections can aggravate congestion in the network.
  1. Using multiple connections can provide a very large overall congestion window. This can be an advantage for TCP implementations that do not support the TCP window scaling extension [JBB92]. However, the aggregate cwnd size across all N connections is equivalent to using a TCP implementation that supports large windows.
  1. The overall cwnd decrease in the face of dropped segments is reduced when using N parallel connections. A single TCP connection reduces the effective size of cwnd to half when a single segment loss is detected. When utilizing N connections each using a window of W bytes, a single drop reduces the window to:
        (N * W) - (W / 2)

Clearly this is a less dramatic reduction in the effective cwnd size than when using a single TCP connection. And, the amount by which the cwnd is decreased is further reduced by increasing N.

The use of multiple data connections can increase the ability of non-SACK TCP implementations to quickly recover from multiple dropped segments without resorting to a timeout, assuming the dropped segments cross connections.

The use of multiple parallel connections makes TCP overly aggressive for many environments and can contribute to congestive collapse in shared networks [FF99]. The advantages provided by using multiple TCP connections are now largely provided by TCP extensions (larger windows, SACKs, etc.). Therefore, the use of a single TCP connection is more "network friendly" than using multiple parallel connections. However, using multiple parallel TCP connections may provide performance improvement in private networks.

3.5.2 Research

Research on the use of multiple parallel TCP connections shows improved performance [IL92,Hah94,AOK95,AKO96]. In addition, research has shown that multiple TCP connections can outperform a single modern TCP connection (with large windows and SACK) [AHKO97]. However, these studies did not consider the impact of using multiple TCP connections on competing traffic. [FF99] argues that using multiple simultaneous connections to transfer a given file may lead to congestive collapse in shared networks.

3.5.3 Implementation Issues

To utilize multiple parallel TCP connections a client application and the corresponding server must be customized. As outlined in [FF99] using multiple parallel TCP connections is not safe (from a congestion control perspective) in shared networks and should not be used.

3.5.4 Topological Considerations

As stated above, [FF99] outlines that the use of multiple parallel connections in a shared network, such as the Internet, may lead to congestive collapse. However, the use of multiple connections may be safe and beneficial in private networks. The specific topology being used will dictate the number of parallel connections required. Some work has been done to determine the appropriate number of connections on the fly [AKO96], but such a mechanism is far from complete.

3.5.5 Possible Interaction and Relationships with Other Research

Using multiple concurrent TCP connections enables use of a large congestion window, much like the TCP window scaling option [JBB92]. In addition, a larger initial congestion window is achieved, similar to using [AFP98] or TCB sharing (see section 3.8).

3.6 Pacing TCP Segments

3.6.1 Mitigation Description

Slow-start takes several round trips to fully open the TCP congestion window over routes with high bandwidth-delay products. For short TCP connections (such as WWW traffic with HTTP/1.0), the slow-start overhead can preclude effective use of the high-bandwidth satellite links. When senders implement slow-start restart after a TCP connection goes idle (suggested by Jacobson and Karels [JK92]), performance is reduced in long-lived (but bursty) connections (such as HTTP/1.1, which uses persistent TCP connections to transfer multiple WWW page elements) [Hei97a].

Rate-based pacing (RBP) is a technique, used in the absence of incoming ACKs, where the data sender temporarily paces TCP segments at a given rate to restart the ACK clock. Upon receipt of the first ACK, pacing is discontinued and normal TCP ACK clocking resumes. The pacing rate may either be known from recent traffic estimates (when restarting an idle connection or from recent prior connections), or may be known through external means (perhaps in a point-to-point or point-to-multipoint satellite network where available bandwidth can be assumed to be large).

In addition, pacing data during the first RTT of a transfer may allow TCP to make effective use of high bandwidth-delay links even for short transfers. However, in order to pace segments during the first RTT a TCP will have to be using a non-standard initial congestion window and a new mechanism to pace outgoing segments rather than send them back-to-back. Determining an appropriate size for the initial cwnd is an open research question. Pacing can also be used to reduce bursts in general (due to buggy TCPs or byte counting, see section 3.2.2 for a discussion on byte counting).

3.6.2 Research

Simulation studies of rate-paced pacing for WWW-like traffic have shown reductions in router congestion and drop rates [VH97a]. In this environment, RBP substantially improves performance compared to slow-start-after-idle for intermittent senders, and it slightly improves performance over burst-full-cwnd-after-idle (because of drops) [VH98]. More recently, pacing has been suggested to eliminate burstiness in networks with ACK filtering [BPK97].

3.6.3 Implementation Issues

RBP requires only sender-side changes to TCP. Prototype implementations of RBP are available [VH97b]. RBP requires an additional sender timer for pacing. The overhead of timer-driven data transfer is often considered too high for practical use. Preliminary experiments suggest that in RBP this overhead is minimal because RBP only requires this timer for one RTT of transmission [VH98]. RBP is expected to make TCP more conservative in sending bursts of data after an idle period in hosts that do not revert to slow start after an idle period. On the other hand, RBP makes TCP more aggressive if the sender uses the slow start algorithm to start the ACK clock after a long idle period.

3.6.4 Topology Considerations

RBP could be used to restart idle TCP connections for all topologies in Section 2. Use at the beginning of new connections would be restricted to topologies where available bandwidth can be estimated out-of-band.

3.6.5 Possible Interaction and Relationships with Other Research

Pacing segments may benefit from sharing state amongst various flows between two hosts, due to the time required to determine the needed information. Additionally, pacing segments, rather than sending back-to-back segments, may make estimating the available bandwidth (as outlined in section 3.2.4) more difficult.

3.7 TCP Header Compression

The TCP and IP header information needed to reliably deliver packets to a remote site across the Internet can add significant overhead, especially for interactive applications. Telnet packets, for example, typically carry only a few bytes of data per packet, and standard IPv4/TCP headers add at least 40 bytes to this; IPv6/TCP headers add at least 60 bytes. Much of this information remains relatively constant over the course of a session and so can be replaced by a short session identifier.

3.7.1 Mitigation Description

Many fields in the TCP and IP headers either remain constant during the course of a session, change very infrequently, or can be inferred from other sources. For example, the source and destination addresses, as well as the IP version, protocol, and port fields generally do not change during a session. Packet length can be deduced from the length field of the underlying link layer protocol provided that the link layer packet is not padded. Packet sequence numbers in a forward data stream generally change with every packet, but increase in a predictable manner.

The TCP/IP header compression methods described in [DNP99,DENP97,Jac90] reduce the overhead of TCP sessions by replacing the data in the TCP and IP headers that remains constant, changes slowly, or changes in a predictable manner with a short "connection number". Using this method, the sender first sends a full TCP/IP header, including in it a connection number that the sender will use to reference the connection. The receiver stores the full header and uses it as a template, filling in some fields from the limited information contained in later, compressed headers. This compression can reduce the size of an IPv4/TCP headers from 40 to as few as 3 to 5 bytes (3 bytes for some common cases, 5 bytes in general).

Compression and decompression generally happen below the IP layer, at the end-points of a given physical link (such as at two routers connected by a serial line). The hosts on either side of the physical link must maintain some state about the TCP connections that are using the link.

The decompresser must pass complete, uncompressed packets to the IP layer. Thus header compression is transparent to routing, for example, since an incoming packet with compressed headers is expanded before being passed to the IP layer.

A variety of methods can be used by the compressor/decompressor to negotiate the use of header compression. For example, the PPP serial line protocol allows for an option exchange, during which time the compressor/decompressor agree on whether or not to use header compression. For older SLIP implementations, [Jac90] describes a mechanism that uses the first bit in the IP packet as a flag.

The reduction in overhead is especially useful when the link is bandwidth-limited such as terrestrial wireless and mobile satellite links, where the overhead associated with transmitting the header bits is nontrivial. Header compression has the added advantage that for the case of uniformly distributed bit errors, compressing TCP/IP headers can provide a better quality of service by decreasing the packet error probability. The shorter, compressed packets are less likely to be corrupted, and the reduction in errors increases the connection's throughput.

Extra space is saved by encoding changes in fields that change relatively slowly by sending only their difference from their values in the previous packet instead of their absolute values. In order to decode headers compressed this way, the receiver keeps a copy of each full, reconstructed TCP header after it is decoded, and applies the delta values from the next decoded compressed header to the reconstructed full header template.

A disadvantage to using this delta encoding scheme where values are encoded as deltas from their values in the previous packet is that if a single compressed packet is lost, subsequent packets with compressed headers can become garbled if they contain fields which depend on the lost packet. Consider a forward data stream of packets with compressed headers and increasing sequence numbers. If packet N is lost, the full header of packet N+1 will be reconstructed at the receiver using packet N-1's full header as a template. Thus the sequence number, which should have been calculated from packet N's header, will be wrong, the checksum will fail, and the packet will be discarded. When the sending TCP times out and retransmits a packet with a full header is forwarded to re-synchronize the decompresser.

It is important to note that the compressor does not maintain any timers, nor does the decompresser know when an error occurred (only the receiving TCP knows this, when the TCP checksum fails). A single bit error will cause the decompresser to lose sync, and subsequent packets with compressed headers will be dropped by the receiving TCP, since they will all fail the TCP checksum. When this happens, no duplicate acknowledgments will be generated, and the decompresser can only re-synchronize when it receives a packet with an uncompressed header. This means that when header compression is being used, both fast retransmit and selective acknowledgments will not be able correct packets lost on a compressed link. The "twice" algorithm, described below, may be a partial solution to this problem.

[DNP99] and [DENP97] describe TCP/IPv4 and TCP/IPv6 compression algorithms including compressing the various IPv6 extension headers as well as methods for compressing non-TCP streams. [DENP97] also augments TCP header compression by introducing the "twice" algorithm. If a particular packet fails to decompress properly, the twice algorithm modifies its assumptions about the inferred fields in the compressed header, assuming that a packet identical to the current one was dropped between the last correctly decoded packet and the current one. Twice then tries to decompress the received packet under the new assumptions and, if the checksum passes, the packet is passed to IP and the decompresser state has been re-synchronized. This procedure can be extended to three or more decoding attempts. Additional robustness can be achieved by caching full copies of packets which don't decompress properly in the hopes that later arrivals will fix the problem. Finally, the performance improvement if the decompresser can explicitly request a full header is
discussed. Simulation results show that twice, in conjunction with the full header request mechanism, can improve throughput over uncompressed streams.

3.7.2 Research

[Jac90] outlines a simple header compression scheme for TCP/IP.

In [DENP97] the authors present the results of simulations showing that header compression is advantageous for both low and medium bandwidth links. Simulations show that the twice algorithm, combined with an explicit header request mechanism, improved throughput by 10-15% over uncompressed sessions across a wide range of bit error rates.

Much of this improvement may have been due to the twice algorithm quickly re-synchronizing the decompresser when a packet is lost. This is because the twice algorithm, applied one or two times when the decompresser becomes unsynchronized, will re-sync the decompresser in between 83% and 99% of the cases examined. This means that packets received correctly after twice has resynchronized the decompresser will cause duplicate acknowledgments. This re- enables the use of both fast retransmit and SACK in conjunction with header compression.

3.7.3 Implementation Issues

Implementing TCP/IP header compression requires changes at both the sending (compressor) and receiving (decompresser) ends of each link that uses compression. The twice algorithm requires very little extra machinery over and above header compression, while the explicit header request mechanism of [DENP97] requires more extensive modifications to the sending and receiving ends of each link that employs header compression. Header compression does not violate TCP's congestion control mechanisms and therefore can be safely implemented in shared networks.

3.7.4 Topology Considerations

TCP/IP header compression is applicable to all of the environments discussed in section 2, but will provide relatively more improvement in situations where packet sizes are small (i.e., overhead is large) and there is medium to low bandwidth and/or higher BER. When TCP's congestion window size is large, implementing the explicit header request mechanism, the twice algorithm, and caching packets which fail to decompress properly becomes more critical.

3.7.5 Possible Interaction and Relationships with Other Research

As discussed above, losing synchronization between a sender and receiver can cause many packet drops. The frequency of losing synchronization and the effectiveness of the twice algorithm may point to using a SACK-based loss recovery algorithm to reduce the impact of multiple lost segments. However, even very robust SACK- based algorithms may not work well if too many segments are lost.

3.8 Sharing TCP State Among Similar Connections

3.8.1 Mitigation Description

Persistent TCP state information can be used to overcome limitations in the configuration of the initial state, and to automatically tune TCP to environments using satellite links and to coordinate multiple TCP connections sharing a satellite link.

TCP includes a variety of parameters, many of which are set to initial values which can severely affect the performance of TCP connections traversing satellite links, even though most TCP parameters are adjusted later after the connection is established. These parameters include initial size of cwnd and initial MSS size. Various suggestions have been made to change these initial conditions, to more effectively support satellite links. However, it is difficult to select any single set of parameters which is effective for all environments.

An alternative to attempting to select these parameters a-priori is sharing state across TCP connections and using this state when initializing a new connection. For example, if all connections to a subnet result in extended congestion windows of 1 megabyte, it is probably more efficient to start new connections with this value, than to rediscover it by requiring the cwnd to increase using slow start over a period of dozens of round-trip times.

3.8.2 Research

Sharing state among connections brings up a number of questions such as what information to share, with whom to share, how to share it, and how to age shared information. First, what information is to be shared must be determined. Some information may be appropriate to share among TCP connections, while some information sharing may be inappropriate or not useful. Next, we need to determine with whom to share information. Sharing may be appropriate for TCP connections sharing a common path to a given host. Information may be shared among connections within a host, or even among connections between different hosts, such as hosts on the same LAN. However, sharing information between connections not traversing the same network may not be appropriate. Given the state to share and the parties that share it, a mechanism for the sharing is required. Simple state, like MSS and RTT, is easy to share, but congestion window information can be shared a variety of ways. The sharing mechanism determines priorities among the sharing connections, and a variety of fairness criteria need to be considered. Also, the mechanisms by which information is aged require further study. See RFC 2140 for a discussion of the security issues in both sharing state within a single host and sharing state among hosts on a subnet. Finally, the security concerns associated with sharing a piece of information need to be carefully considered before introducing such a mechanism. Many of these open research questions must be answered before state sharing can be widely deployed.

The opportunity for such sharing, both among a sequence of connections, as well as among concurrent connections, is described in more detail in [Tou97]. The state management itself is largely an implementation issue, however what information should be shared and the specific ways in which the information should be shared is an open question.

Sharing parts of the TCB state was originally documented in T/TCP [Bra92], and is used there to aggregate RTT values across connection instances, to provide meaningful average RTTs, even though most connections are expected to persist for only one RTT. T/TCP also shares a connection identifier, a sequence number separate from the window number and address/port pairs by which TCP connections are typically distinguished. As a result of this shared state, T/TCP allows a receiver to pass data in the SYN segment to the receiving application, prior to the completion of the three-way handshake, without compromising the integrity of the connection. In effect, this shared state caches a partial handshake from the previous connection, which is a variant of the more general issue of TCB sharing.

Sharing state among connections (including transfers using non-TCP protocols) is further investigated in [BRS99].

3.8.3 Implementation Issues

Sharing TCP state across connections requires changes to the sender's TCP stack, and possibly the receiver's TCP stack (as in the case of T/TCP, for example). Sharing TCP state may make a particular TCP connection more aggressive. However, the aggregate traffic should be more conservative than a group of independent TCP connections. Therefore, sharing TCP state should be safe for use in shared networks. Note that state sharing does not present any new security problems within multiuser hosts. In such a situation, users can steal network resources from one another with or without state
sharing.

3.8.4 Topology Considerations

It is expected that sharing state across TCP connections may be useful in all network environments presented in section 2.

3.8.5 Possible Interaction and Relationships with Other Research

The state sharing outlined above is very similar to the Congestion Manager proposal [BRS99] that attempts to share congestion control information among both TCP and UDP flows between a pair of hosts.

3.9 ACK Congestion Control

In highly asymmetric networks, a low-speed return link can restrict the performance of the data flow on a high-speed forward link by limiting the flow of acknowledgments returned to the data sender. For example, if the data sender uses 1500 byte segments, and the receiver generates 40 byte acknowledgments (IPv4, TCP without options), the reverse link will congest with ACKs for asymmetries of more than 75:1 if delayed ACKs are used, and 37:1 if every segment is acknowledged. For a 1.5 Mb/second data link, ACK congestion will occur for reverse link speeds below 20 kilobits/sec. These levels of asymmetry will readily occur if the reverse link is shared among multiple satellite receivers, as is common in many VSAT satellite networks. If a terrestrial modem link is used as a reverse link, ACK congestion is also likely, especially as the speed of the forward link is increased. Current congestion control mechanisms are aimed at controlling the flow of data segments, but do not affect the flow of ACKs.

In [KVR98] the authors point out that the flow of acknowledgments can be restricted on the low-speed link not only by the bandwidth of the link, but also by the queue length of the router. The router may limit its queue length by counting packets, not bytes, and therefore begin discarding ACKs even if there is enough bandwidth to forward them.

3.9.1 Mitigation Description

ACK Congestion Control extends the concept of flow control for data segments to acknowledgment segments. In the method described in [BPK97], any intermediate router can mark an acknowledgment with an Explicit Congestion Notification (ECN) bit once the queue occupancy in the router exceeds a given threshold. The data sender (which receives the acknowledgment) must "echo" the ECN bit back to the data receiver (see section 3.3.3 for a more detailed discussion of ECN). The proposed algorithm for marking ACK segments with an ECN bit is Random Early Detection (RED) [FJ93]. In response to the receipt of ECN marked data segments, the receiver will dynamically reduce the rate of acknowledgments using a multiplicative backoff. Once segments without ECN are received, the data receiver speeds up acknowledgments using a linear increase, up to a rate of either 1 (no delayed ACKs) or 2 (normal delayed ACKs) data segments per ACK. The authors suggest that an ACK be generated at least once per window, and ideally a few times per window.

As in the RED congestion control mechanism for data flow, the bottleneck gateway can randomly discard acknowledgments, rather than marking them with an ECN bit, once the queue fills beyond a given threshold.

3.9.2 Research

[BPK97] analyze the effect of ACK Congestion Control (ACC) on the performance of an asymmetric network. They note that the use of ACC, and indeed the use of any scheme which reduces the frequency of acknowledgments, has potential unwanted side effects. Since each ACK will acknowledge more than the usual one or two data segments, the likelihood of segment bursts from the data sender is increased. In addition, congestion window growth may be impeded if the receiver grows the window by counting received ACKs, as mandated by [Ste97,APS99]. The authors therefore combine ACC with a series of modifications to the data sender, referred to as TCP Sender Adaptation (SA). SA combines a limit on the number of segments sent in a burst, regardless of window size. In addition, byte counting (as opposed to ACK counting) is employed for window growth. Note that byte counting has been studied elsewhere and can introduce side-effects, as well [All98].

The results presented in [BPK97] indicate that using ACC and SA will reduce the bursts produced by ACK losses in unmodified (Reno) TCP. In cases where these bursts would lead to data loss at an intermediate router, the ACC and SA modification significantly improve the throughput for a single data transfer. The results further suggest that the use of ACC and SA significantly improve fairness between two simultaneous transfers.

ACC is further reported to prevent the increase in round trip time (RTT) that occurs when an unmodified TCP fills the reverse router queue with acknowledgments.

In networks where the forward direction is expected to suffer losses in one of the gateways, due to queue limitations, the authors report at best a very slight improvement in performance for ACC and SA, compared to unmodified Reno TCP.

3.9.3 Implementation Issues

Both ACC and SA require modification of the sending and receiving hosts, as well as the bottleneck gateway. The current research suggests that implementing ACC without the SA modifications results in a data sender which generates potentially disruptive segment bursts. It should be noted that ACC does require host modifications if it is implemented in the way proposed in [BPK97]. The authors note that ACC can be implemented by discarding ACKs (which requires only a gateway modification, but no changes in the hosts), as opposed to marking them with ECN. Such an implementation may, however, produce bursty data senders if it is not combined with a burst mitigation technique. ACC requires changes to the standard ACKing behavior of a receiving TCP and therefore is not recommended for use in shared networks.

3.9.4 Topology Considerations

Neither ACC nor SA require the storage of state in the gateway. These schemes should therefore be applicable for all topologies, provided that the hosts using the satellite or hybrid network can be modified. However, these changes are expected to be especially beneficial to networks containing asymmetric satellite links.

3.9.5 Possible Interaction and Relationships with Other Research

Note that ECN is a pre-condition for using ACK congestion control. Additionally, the ACK Filtering algorithm discussed in the next section attempts to solve the same problem as ACC. Choosing between the two algorithms (or another mechanism) is currently an open research question.

3.10 ACK Filtering

ACK Filtering (AF) is designed to address the same ACK congestion effects described in 3.9. Contrary to ACC, however, AF is designed to operate without host modifications.

3.10.1 Mitigation Description

AF takes advantage of the cumulative acknowledgment structure of TCP. The bottleneck router in the reverse direction (the low speed link) must be modified to implement AF. Upon receipt of a segment which represents a TCP acknowledgment, the router scans the queue for redundant ACKs for the same connection, i.e. ACKs which acknowledge portions of the window which are included in the most recent ACK. All of these "earlier" ACKs are removed from the queue and discarded.

The router does not store state information, but does need to implement the additional processing required to find and remove segments from the queue upon receipt of an ACK.

3.10.2 Research

[BPK97] analyzes the effects of AF. As is the case in ACC, the use of ACK filtering alone would produce significant sender bursts, since the ACKs will be acknowledging more previously-unacknowledged data. The SA modifications described in 3.9.2 could be used to prevent those bursts, at the cost of requiring host modifications. To prevent the need for modifications in the TCP stack, AF is more likely to be paired with the ACK Reconstruction (AR) technique, which can be implemented at the router where segments exit the slow reverse link.

AR inspects ACKs exiting the link, and if it detects large "gaps" in the ACK sequence, it generates additional ACKs to reconstruct an acknowledgment flow which more closely resembles what the data sender would have seen had ACK Filtering not been introduced. AR requires two parameters; one parameter is the desired ACK frequency, while the second controls the spacing, in time, between the release of consecutive reconstructed ACKs.

In [BPK97], the authors show the combination of AF and AR to increase throughput, in the networks studied, over both unmodified TCP and the ACC/SA modifications. Their results also strongly suggest that the use of AF alone, in networks where congestion losses are expected, decreases performance (even below the level of unmodified TCP Reno) due to sender bursting.

AF delays acknowledgments from arriving at the receiver by dropping earlier ACKs in favor of later ACKs. This process can cause a slight hiccup in the transmission of new data by the TCP sender.

3.10.3 Implementation Issues

Both ACK Filtering and ACK Reconstruction require only router modification. However, the implementation of AR requires some storage of state information in the exit router. While AF does not require storage of state information, its use without AR (or SA) could produce undesired side effects. Furthermore, more research is required regarding appropriate ranges for the parameters needed in
AR.

3.10.4 Topology Considerations

AF and AR appear applicable to all topologies, assuming that the storage of state information in AR does not prove to be prohibitive for routers which handle large numbers of flows. The fact that TCP stack modifications are not required for AF/AR makes this approach attractive for hybrid networks and networks with diverse types of hosts. These modifications, however, are expected to be most beneficial in asymmetric network paths.

On the other hand, the implementation of AF/AR requires the routers to examine the TCP header, which prohibits their use in secure networks where IPSEC is deployed. In such networks, AF/AR can be effective only inside the security perimeter of a private, or virtual private network, or in private networks where the satellite link is protected only by link-layer encryption (as opposed to IPSEC). ACK Filtering is safe to use in shared networks (from a congestion control point-of-view), as the number of ACKs can only be reduced, which makes TCP less aggressive. However, note that while TCP is less aggressive, the delays that AF induces (outlined above) can lead to larger bursts than would otherwise occur.

3.10.5 Possible Interaction and Relationships with Other Research

ACK Filtering attempts to solve the same problem as ACK Congestion Control (as outlined in section 3.9). Which of the two algorithms is more appropriate is currently an open research question.

4 Conclusions

This document outlines TCP items that may be able to mitigate the performance problems associated with using TCP in networks containing satellite links. These mitigations are not IETF standards track mechanisms and require more study before being recommended by the IETF. The research community is encouraged to examine the above mitigations in an effort to determine which are safe for use in shared networks such as the Internet.

5 Security Considerations

Several of the above sections noted specific security concerns which a given mitigation aggravates.

Additionally, any form of wireless communication link is more susceptible to eavesdropping security attacks than standard wire- based links due to the relative ease with which an attacker can watch the network and the difficultly in finding attackers monitoring the network.

6 Acknowledgments

Our thanks to Aaron Falk and Sally Floyd, who provided very helpful comments on drafts of this document.

7 References

   [AFP98]   Allman, M., Floyd, S. and C. Partridge, "Increasing TCP's
   
             Initial Window", RFC 2414, September 1998.
   
   [AGS99]   Allman, M., Glover, D. and L. Sanchez, "Enhancing TCP Over
             Satellite Channels using Standard Mechanisms", BCP 28, RFC
             2488, January 1999.
   
   [AHKO97]  Mark Allman, Chris Hayes, Hans Kruse, Shawn Ostermann.  TCP
             Performance Over Satellite Links.  In Proceedings of the
             5th International Conference on Telecommunication Systems,
             March 1997.
   
   [AHO98]   Mark Allman, Chris Hayes, Shawn Ostermann.  An Evaluation
             of TCP with Larger Initial Windows.  Computer Communication
             Review, 28(3), July 1998.
   
   [AKO96]   Mark Allman, Hans Kruse, Shawn Ostermann.  An Application-
             Level Solution to TCP's Satellite Inefficiencies.  In
             Proceedings of the First International Workshop on
             Satellite-based Information Services (WOSBIS), November
             1996.
   
   [All97a]  Mark Allman.  Improving TCP Performance Over Satellite
             Channels.  Master's thesis, Ohio University, June 1997.
   
   [All97b]  Mark Allman.  Fixing Two BSD TCP Bugs.  Technical Report
             CR-204151, NASA Lewis Research Center, October 1997.
   
   [All98]   Mark Allman. On the Generation and Use of TCP
             Acknowledgments.  ACM Computer Communication Review, 28(5),
             October 1998.
   
   [AOK95]   Mark Allman, Shawn Ostermann, Hans Kruse.  Data Transfer
             Efficiency Over Satellite Circuits Using a Multi-Socket
             Extension to the File Transfer Protocol (FTP).  In
             Proceedings of the ACTS Results Conference, NASA Lewis
             Research Center, September 1995.
   
   [AP99]    Mark Allman, Vern Paxson.  On Estimating End-to-End Network
             Path Properties. ACM SIGCOMM, September 1999.
   
   [APS99]   Allman, M., Paxson, V. and W. Richard Stevens, "TCP
             Congestion Control", RFC 2581, April 1999.
   
   [BCC+98]  Braden, B., Clark, D., Crowcroft, J., Davie, B., Deering,
             S., Estrin, D., Floyd, S., Jacobson, V., Minshall, G.,
             Partridge, C., Peterson, L., Ramakrishnan, K., Shenker, S.,
             Wroclawski, J. and L. Zhang, "Recommendations on Queue
             Management and Congestion Avoidance in the Internet", RFC
             2309, April 1998.
   
   [BKVP97]  B. Bakshi and P. Krishna and N. Vaidya and D. Pradham,
             "Improving Performance of TCP over Wireless Networks", 17th
             International Conference on Distributed Computing Systems
             (ICDCS), May 1997.
   
   [BPK97]   Hari Balakrishnan, Venkata N. Padmanabhan, and Randy H.
             Katz.  The Effects of Asymmetry on TCP Performance.  In
             Proceedings of the ACM/IEEE Mobicom, Budapest, Hungary,
             ACM.  September, 1997.
   
   [BPK98]   Hari Balakrishnan, Venkata Padmanabhan, Randy H. Katz.  The
             Effects of Asymmetry on TCP Performance.  ACM Mobile
             Networks and Applications (MONET), 1998 (to appear).
   
   [BPSK96]  H. Balakrishnan and V. Padmanabhan and S. Sechan and R.
             Katz, "A Comparison of Mechanisms for Improving TCP
             Performance over Wireless Links", ACM SIGCOMM, August 1996.
   
   [Bra89]   Braden, R., "Requirements for Internet Hosts --
             Communication Layers", STD 3, RFC 1122, October 1989.
   
   [Bra92]   Braden, R., "Transaction TCP -- Concepts", RFC 1379,
             September 1992.
   
   [Bra94]   Braden, R., "T/TCP -- TCP Extensions for Transactions:
             Functional Specification", RFC 1644, July 1994.
   
   [BRS99]   Hari Balakrishnan, Hariharan Rahul, and Srinivasan Seshan.
             An Integrated Congestion Management Architecture for
             Internet Hosts.  ACM SIGCOMM, September 1999.
   
   [ddKI99]  M. deVivo, G.O. deVivo, R. Koeneke, G. Isern.  Internet
             Vulnerabilities Related to TCP/IP and T/TCP.  Computer
             Communication Review, 29(1), January 1999.
   
   [DENP97]  Mikael Degermark, Mathias Engan, Bjorn Nordgren, Stephen
             Pink.  Low-Loss TCP/IP Header Compression for Wireless
             Networks.  ACM/Baltzer Journal on Wireless Networks, vol.3,
             no.5, p. 375-87.
   
   [DMT96]   R. C. Durst and G. J. Miller and E. J. Travis, "TCP
             Extensions for Space Communications", Mobicom 96, ACM, USA,
             1996.
   
   [DNP99]   Degermark, M., Nordgren, B. and S. Pink, "IP Header
             Compression", RFC 2507, February 1999.
   
   [FF96]    Kevin Fall, Sally Floyd.  Simulation-based Comparisons of
             Tahoe, Reno, and SACK TCP.  Computer Communication Review,
             V. 26 N. 3, July 1996, pp. 5-21.
   
   [FF99]    Sally Floyd, Kevin Fall.  Promoting the Use of End-to-End
             Congestion Control in the Internet, IEEE/ACM Transactions
             on Networking, August 1999.
   
   [FH99]    Floyd, S. and T. Henderson, "The NewReno Modification to
             TCP's Fast Recovery Algorithm", RFC 2582, April 1999.
   
   [FJ93]    Sally Floyd and Van Jacobson.  Random Early Detection
             Gateways for Congestion Avoidance, IEEE/ACM Transactions on
             Networking, V. 1 N. 4, August 1993.
   
   [Flo91]   Sally Floyd.  Connections with Multiple Congested Gateways
             in Packet-Switched Networks, Part 1: One-way Traffic.  ACM
             Computer Communications Review, V. 21, N. 5, October 1991.
   
   [Flo94]   Sally Floyd.  TCP and Explicit Congestion Notification, ACM
             Computer Communication Review, V. 24 N. 5, October 1994.
   
   [Flo99]   Sally Floyd.  "Re: TCP and out-of-order delivery", email to
             end2end-interest mailing list, February, 1999.
   
   [Hah94]   Jonathan Hahn.  MFTP: Recent Enhancements and Performance
             Measurements.  Technical Report RND-94-006, NASA Ames
             Research Center, June 1994.
   
   [Hay97]   Chris Hayes.  Analyzing the Performance of New TCP
             Extensions Over Satellite Links.  Master's Thesis, Ohio
             University, August 1997.
   
   [HK98]    Tom Henderson, Randy Katz.  On Improving the Fairness of
             TCP Congestion Avoidance.  Proceedings of IEEE Globecom `98
             Conference, 1998.
   
   [HK99]    Tim Henderson, Randy Katz.  Transport Protocols for
             Internet-Compatible Satellite Networks, IEEE Journal on
             Selected Areas of Communications, February, 1999.
   
   [Hoe95]   J. Hoe, Startup Dynamics of TCP's Congestion Control and
             Avoidance Schemes. Master's Thesis, MIT, 1995.
   
   [Hoe96]   Janey Hoe.  Improving the Startup Behavior of a Congestion
             Control Scheme for TCP.  In ACM SIGCOMM, August 1996.
   
   [IL92]    David Iannucci and John Lakashman.  MFTP: Virtual TCP
             Window Scaling Using Multiple Connections.  Technical
             Report RND-92-002, NASA Ames Research Center, January 1992.
   
   [Jac88]   Van Jacobson.  Congestion Avoidance and Control.  In
             Proceedings of the SIGCOMM '88, ACM.  August, 1988.
   
   [Jac90]   Jacobson, V., "Compressing TCP/IP Headers", RFC 1144,
             February 1990.
   
   [JBB92]   Jacobson, V., Braden, R. and D. Borman, "TCP Extensions for
             High Performance", RFC 1323, May 1992.
   
   [JK92]    Van Jacobson and Mike Karels.  Congestion Avoidance and
             Control.  Originally appearing in the proceedings of
             SIGCOMM '88 by Jacobson only, this revised version includes
             an additional appendix.  The revised version is available
             at ftp://ftp.ee.lbl.gov/papers/congavoid.ps.Z.  1992.
   
   [Joh95]   Stacy Johnson.  Increasing TCP Throughput by Using an
             Extended Acknowledgment Interval.  Master's Thesis, Ohio
             University, June 1995.
   
   [KAGT98]  Hans Kruse, Mark Allman, Jim Griner, Diepchi Tran.  HTTP
             Page Transfer Rates Over Geo-Stationary Satellite Links.
             March 1998. Proceedings of the Sixth International
             Conference on Telecommunication Systems.
   
   [Kes91]   Srinivasan Keshav.  A Control Theoretic Approach to Flow
             Control.  In ACM SIGCOMM, September 1991.
   
   [KM97]    S. Keshav, S. Morgan. SMART Retransmission: Performance
             with Overload and Random Losses. Proceeding of Infocom.
             1997.
   
   [KVR98]   Lampros Kalampoukas, Anujan Varma, and K. K.Ramakrishnan.
             Improving TCP Throughput Over Two-Way Asymmetric Links:
             Analysis and Solutions.  Measurement and Modeling of
             Computer Systems, 1998, Pages 78-89.
   
   [MM96a]   M. Mathis, J. Mahdavi, "Forward Acknowledgment: Refining
             TCP Congestion Control," Proceedings of SIGCOMM'96, August,
             1996, Stanford, CA.  Available from
             http://www.psc.edu/networking/papers/papers.html
   
   [MM96b]   M. Mathis, J. Mahdavi, "TCP Rate-Halving with Bounding
             Parameters" Available from
             http://www.psc.edu/networking/papers/FACKnotes/current.
   
   [MMFR96]  Mathis, M., Mahdavi, J., Floyd, S. and A. Romanow, "TCP
             Selective Acknowledgment Options", RFC 2018, October 1996.
   
   [MSMO97]  M. Mathis, J. Semke, J. Mahdavi, T. Ott, "The Macroscopic
             Behavior of the TCP Congestion Avoidance
             Algorithm",Computer Communication Review, volume 27,
             number3, July 1997.  Available from
             http://www.psc.edu/networking/papers/papers.html
   
   [MV98]    Miten N. Mehta and Nitin H. Vaidya.  Delayed Duplicate-
             Acknowledgments: A Proposal to Improve Performance of TCP
             on Wireless Links.  Technical Report 98-006, Department of
             Computer Science, Texas A&M University, February 1998.
   
   [Nic97]   Kathleen Nichols.  Improving Network Simulation with
             Feedback.  Com21, Inc. Technical Report.  Available from
             http://www.com21.com/pages/papers/068.pdf.

[PADHV99] Paxson, V., Allman, M., Dawson, S., Heavens, I. and B.

Volz, "Known TCP Implementation Problems", RFC 2525, March 1999.

   [Pax97]   Vern Paxson.  Automated Packet Trace Analysis of TCP
             Implementations.  In Proceedings of ACM SIGCOMM, September
             1997.
   
   [PN98]    Poduri, K. and K. Nichols, "Simulation Studies of Increased
             Initial TCP Window Size", RFC 2415, September 1998.
   
   [Pos81]   Postel, J., "Transmission Control Protocol", STD 7, RFC
             793, September 1981.
   
   [RF99]    Ramakrishnan, K. and S. Floyd, "A Proposal to add Explicit
             Congestion Notification (ECN) to IP", RFC 2481, January
             1999.
   
   [SF98]    Nihal K. G. Samaraweera and Godred Fairhurst,
             "Reinforcement of TCP error Recovery for Wireless
             Communication", Computer Communication Review, volume 28,
             number 2, April 1998.
   
   [SP98]    Shepard, T. and C. Partridge, "When TCP Starts Up With Four
             Packets Into Only Three Buffers", RFC 2416, September 1998.
   
   [Ste97]   Stevens, W., "TCP Slow Start, Congestion Avoidance, Fast
             Retransmit, and Fast Recovery Algorithms", RFC 2001,
             January 1997.
   
   [Sut98]   B. Suter, T. Lakshman, D. Stiliadis, and A. Choudhury.
             Design Considerations for Supporting TCP with Per-flow
             Queueing.  Proceedings of IEEE Infocom `98 Conference,
             1998.
   
   [Tou97]   Touch, J., "TCP Control Block Interdependence", RFC 2140,
             April 1997.
   
   [VH97a]   Vikram Visweswaraiah and John Heidemann.  Improving Restart
             of Idle TCP Connections.  Technical Report 97-661,
             University of Southern California, 1997.
   
   [VH97b]   Vikram Visweswaraiah and John Heidemann.  Rate-based pacing
             Source Code Distribution, Web page:
             http://www.isi.edu/lsam/publications/rate_based_pacing/README.html
             November, 1997.
   
   [VH98]    Vikram Visweswaraiah and John Heidemann.  Improving Restart
             of Idle TCP Connections (revised).  Submitted for
             publication.

8 Authors' Addresses

   Mark Allman
   NASA Glenn Research Center/BBN Technologies
   Lewis Field
   21000 Brookpark Rd.  MS 54-2
   Cleveland, OH  44135
   
   EMail: mallman@grc.nasa.gov
   http://roland.grc.nasa.gov/~mallman
   
   Spencer Dawkins
   Nortel
   P.O.Box 833805
   Richardson, TX 75083-3805
   
   EMail: Spencer.Dawkins.sdawkins@nt.com
   
   Dan Glover
   NASA Glenn Research Center
   Lewis Field
   21000 Brookpark Rd.  MS 3-6
   Cleveland, OH  44135
   
   EMail: Daniel.R.Glover@grc.nasa.gov
   http://roland.grc.nasa.gov/~dglover
   
   Jim Griner
   NASA Glenn Research Center
   Lewis Field
   21000 Brookpark Rd.  MS 54-2
   Cleveland, OH  44135
   
   EMail: jgriner@grc.nasa.gov
   http://roland.grc.nasa.gov/~jgriner
   
   Diepchi Tran
   NASA Glenn Research Center
   Lewis Field
   21000 Brookpark Rd.  MS 54-2
   Cleveland, OH  44135
   
   EMail: dtran@grc.nasa.gov

Tom Henderson
University of California at Berkeley
Phone: +1 (510) 642-8919

   EMail: tomh@cs.berkeley.edu
   URL: http://www.cs.berkeley.edu/~tomh/

John Heidemann
University of Southern California/Information Sciences Institute 4676 Admiralty Way
Marina del Rey, CA 90292-6695

EMail:

          johnh@isi.edu

Joe Touch
University of Southern California/Information Sciences Institute 4676 Admiralty Way
Marina del Rey, CA 90292-6601
USA

   Phone: +1 310-448-9151
   Fax:   +1 310-823-6714
   URL:   http://www.isi.edu/touch
   EMail: touch@isi.edu

Hans Kruse
J. Warren McClure School of Communication Systems Management Ohio University
9 S. College Street
Athens, OH 45701

   Phone: 740-593-4891
   Fax: 740-593-4889
   EMail: hkruse1@ohiou.edu
   http://www.csm.ohiou.edu/kruse
   
   Shawn Ostermann
   School of Electrical Engineering and Computer Science
   Ohio University
   416 Morton Hall
   Athens, OH  45701
   
   Phone: (740) 593-1234
   EMail: ostermann@cs.ohiou.edu

Keith Scott
The MITRE Corporation
M/S W650
1820 Dolley Madison Blvd.
McLean VA 22102-3481

EMail:

          kscott@mitre.org
   
   Jeffrey Semke
   Pittsburgh Supercomputing Center
   4400 Fifth Ave.
   Pittsburgh, PA  15213
   
   EMail: semke@psc.edu
   http://www.psc.edu/~semke

9 Full Copyright Statement

Copyright © The Internet Society (2000). All Rights Reserved.

This document and translations of it may be copied and furnished to others, and derivative works that comment on or otherwise explain it or assist in its implementation may be prepared, copied, published and distributed, in whole or in part, without restriction of any kind, provided that the above copyright notice and this paragraph are included on all such copies and derivative works. However, this document itself may not be modified in any way, such as by removing the copyright notice or references to the Internet Society or other Internet organizations, except as needed for the purpose of developing Internet standards in which case the procedures for copyrights defined in the Internet Standards process must be followed, or as required to translate it into languages other than English.

The limited permissions granted above are perpetual and will not be revoked by the Internet Society or its successors or assigns.

This document and the information contained herein is provided on an "AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING TASK FORCE DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.

Acknowledgement

Funding for the RFC Editor function is currently provided by the Internet Society.