TCP Congestion Window Size Research Articles

Accurate mathematical modeling and solution of TCP congestion window size distribution

The congestion window size distribution is important in TCP as it reveals the statistical variation of the sending rate. It also quantifies the network congestion conditions. A semi-Markov model is used to develop an accurate mathematical framework for solving for TCP congestion window size. The new model improves on the accuracy of a widely accepted mathematical model in the literature for the congestion window size distribution. The correct statistics of a sum of weighted exponential random variables are identified and incorporated into the new mathematical model. Simulation results from both Matlab and NS2 demonstrate the accuracy of the new model. In one example, the new model predicts that 97.5% of congestion windows will have length less than 82 whereas the previous accepted model predicts 97.5% of the windows will have sizes less than 72, and hence the previous model can lead to buffer and caching requirement estimates that are too small when designing practical networks.

Estimating TCP flow completion time distributions

The flow completion time is an important performance metric for network users, and it is used in designing scheduling and congestion control algorithms. The existing TCP flow completion time distributions are calculated from complicated models, and they are only suitable for short TCP flows. This paper is motivated to build a simple model to derive TCP flow completion time distribution that is suitable for both short and long TCP flows. To build such a model, the distribution of the packet transmission latency under TCP congestion control is firstly calculated. The packet transmission latency is the time needed to successfully deliver a packet, including the time for retransmissions due to packet losses. Semi-Markov process is used to model the TCP congestion window evolution. Using the packet transmission latency distribution together with the TCP congestion window size information, a new parallel queueing model which describes how a file is transmitted by a TCP flow is formed. In the parallel queueing model, the flow completion time distribution is derived, and simulation results from NS2 demonstrate the accuracy of the proposed analysis.

ENGLISH

A mobile ad hoc network (MANET) is determined as a network that has many free autonomous nodes i.e. that can move locations and configure itself. It is often composed of mobile devices that can arrange themselves in various ways using wireless connections such as a cellular or satellite transmission to connect to various networks and operate without strict top-down network or any centralized administration. There are various applications of MANET, In Commercial Sector: Ad hoc can be implemented in emergency, rescue operations for disaster relief efforts, e.g. in fire, flood, or earth-quake. Local Level: e.g. conference or classroom. In mobile ad hoc network, congestion can take place between two intermediate nodes, Congestion occurs when bandwidth is insufficient and network data traffic exceeds capacity. This paper mainly studies and compares the performance of MANET routing protocols namely PSR, DSR, DSDV and AODV under various traffic loads with various maximum TCP congestion window size to improve congestion control in the routing. The metrics used to compare routing protocol performance are packet delivery ratio, average routing load, the average end-to-end time required with delay and average network throughput and mainly to overcome congestion situation and avoid packet loss in wireless networks.

위성 통신에서 Cross-layer 기반 PEP 성능 평가

위성 통신은 넓은 지역에 통신 서비스를 제공하는 광역 네트워크이다. 하지만, 위성 통신은 대역폭이 제한되고 매우 긴 전파 지연 시간과 높은 BER(Bit Error Rate)로 인해 위성 통신을 고려하지 않은 TCP의 성능이 저하 된다. 본 논문에서는 위성 통신에서의 TCP 성능을 향상시키기 위해 DVB-RCS(Digital Video Broadcasting-Return Channel via Satellite) 네트워크 환경을 고려한 cross-layer 기반 PEP(Performance Enhancing Proxy) 기술을 제안한다. 제안하는 프로토콜은 TCP와 링크 계층 간 위성 자원 할당 정보를 정보 교환을 통해 최적의 TCP CWND(Congestion Window)를 설정한다. 제안한 프로토콜의 성능 평가를 위해 리눅스 기반의 PEP 테스트 베드를 구현하였다. 성능 평가 결과 다양한 BER에서 제안한 프로토콜은 위성 자원 할당 정보를 이용하여 최적의 TCP CWND 크기를 설정하기 때문에 단일 및 다중 세션 환경에서 기존 TCP 보다 제안한 프로토콜이 더 좋은 성능을 보여준다는 것을 확인할 수 있었다. Satellite communication is a wide area network (WAN) which provides communication service worldwide. However, the performance of TCP can be seriously degraded in the satellite networks due to limited bandwidth, long round-trip time (RTT) and high bit error rate (BER) over satellite links. In order to improve the performance of TCP, this paper proposes cross-layer Performance Enhancing Proxy (PEP) in digital video broadcasting-return channel via satellite (DVB-RCS) networks. The proposed protocol sets TCP Congestion Window (CWND) size by using satellite resource allocation information exchanged between TCP and the link-layer. we implement PEP testbed based on Linux to evaluate the performance of the proposed protocol. The simulation results show that the proposed protocol performs better than standard TCP both in single and multiple sessions in variant BER, because the proposed protocol sets TCP CWND size by using satellite resource allocation.

Analysis of TCP-unfairness from MAC Layer Perspective in Wireless Ad-hoc Networks

In Ad-hoc networks every single node or station takes the responsibility to forward the data packets from the nodes that lie in the range with them. This additional routing responsibility might lead to performance degradation in some cases. Moreover, it is also reported that TCP which is designed for wired networks exhibit poor performance in terms of throughput and fairness in ad-hoc networks. Attempts to alleviate this issue included solutions from both TCP and MAC layer protocol modifications. This work aims at in-depth analysis of MAC level solutions (basic 802.11, 802.11 with RTS/CTS and Collision Detection Mechanism Based-MAC) to address TCP unfairness problem in 802.11-based multi-hop networks. A set of simulation based experiments are conducted and their observations are analyzed using metrics like throughput, goodput and loss probability. The metric loss probability is taken to investigate the protocols both from the perspective of congestion and collision i.e. the impact of congestion window and contention procedure on the end-to-end performance. The results obtained from the simulation revealed the inefficiency of the CDMB-MAC in the scenarios where hidden and exposed terminals do not exist. Further when TCP congestion window is set to 1 the basic 802.11 MAC is found sufficient to provide higher throughput with better fairness among the TCP connections but when TCP congestion window size is increased to 32 the 802.11 MAC with RTS/CTS scheme gives better performance. Added to this all the three MAC protocols fail in the scenario where the source and destination of a connection lie in the transmission ranges of destination and source of other connections. Therefore it can be concluded that it is indispensable to design an adaptive MAC protocol that could provide Fairness among TCP connections in all scenarios.

A Quality-Centric TCP-Friendly Congestion Control for Multimedia Transmission

In this paper, we propose a quality-centric congestion control for multimedia streaming over wired IP networks, which we refer to as media-TCP-friendly congestion control (MTCC). Our solution adapts the sending rate to both the network condition and the application characteristics by explicitly considering the distortion impacts, delay deadlines, and interdependencies of different video packet classes. The media-aware solution is able to provide differential services for transmitting various packet classes and thereby, further improves the multimedia streaming quality compared to the conventional network-aware congestion control. We use finite-horizon Markov decision process (FHMDP) to determine the optimal congestion control policy that maximizes the long-term multimedia quality, while adhering to the horizon- <formula formulatype="inline" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex Notation="TeX">$K$</tex></formula> TCP-friendliness constraint, which ensures long-term fairness with existing TCP applications. Moreover, the proposed MTCC is able to achieve quality-based fairness among multimedia users. We derive sufficient conditions for multiple multimedia users to achieve quality-based fairness using MTCC congestion control. Note that the proposed solution only modifies the adaptation mechanism of the TCP congestion window size at the sender, without changing the design at the receiver side (i.e., each current TCP receiver can correctly receive and process MTCC streams). Our simulation results show that MTCC achieves more than 3 dB improvement in terms of PSNR over the conventional TCP congestion control approaches, with the largest improvements observed for real-time streaming applications requiring stringent playback delays.

Modelling and Analysis of TCP Performance in Wireless Multihop Networks

Researchers have used extensive simulation and experimental studies to understand TCP performance in wireless multihop networks. In contrast, the objective of this paper is to theoretically analyze TCP performance in this environment. By examining the case of running one TCP session over a string topology, a system model for analyzing TCP performance in multihop wireless networks is proposed, which considers packet buffering, contention of nodes for access to the wireless channel, and spatial reuse of the wireless channel. Markov chain modelling is applied to analyze this system model. Analytical results show that when the number of hops that the TCP session crosses is ?xed, the TCP throughput is independent of the TCP congestion window size. When the number of hops increases from one, the TCP throughput decreases ?rst, and then stabilizes when the number of hops becomes large. The analysis is validated by comparing the numerical and simulation results.

Opportunistic link scheduling for multihop wireless networks

This paper studies throughput improvement for TCP traffic in IEEE 802.11-based multihop ad hoc wireless networks. Due to the incompatibility between TCP and the IEEE 802.11 distributed coordination function (DCF) protocols, the reaction of TCP in case of packet losses can significantly reduce TCP end-to-end throughput. In this paper, we propose an opportunistic link scheduling (OLS), which is a simple enhancement to the IEEE 802.11 DCF protocol and intends to improve the compatibility between TCP and MAC layer protocols in multihop ad hoc networks. With OLS, a link with a good channel condition is allowed to transmit multiple packets consecutively as a burst, while the burst size depends on both physical channel fading and MAC layer collisions. The protocol also includes a mechanism to prevent starvation of nodes with poor channel conditions. An analytical model is developed for a four-hop chain to study the effect of the burst size and TCP congestion window size on the end-to-end transmission throughput in opportunistic link scheduling. Our results show that OLS can significantly improve the end-to-end transmission throughput, while keeping reasonably low transmission delay. The protocol is easy to implement and requires only slight modifications to the IEEE 802.11 protocol.

TCP Performance Dynamics and Link-Layer Adaptation Based Optimization Methods for Wireless Networks

Almost a decade long research on the performance of TCP in wireless networks has resulted in many proposals and solutions to the problem of TCP throughput degradation. Several of these measures, however, have their share of drawbacks. With the continuing emergence of wireless technologies ever since the work on TCP performance over wireless began, smart link-layer mechanisms like adaptive modulation and coding, power control, and incremental redundancy have been designed and deployed. In this work, we outline a cross-layer optimization framework based on the congestion control dynamics of a bulk-transfer TCP flow and demonstrate its application to networks which offer link-layer adaptive measures. We begin by observing that the TCP's congestion window dynamics are comprised of certain recurring patterns which we term as cycles. We then overlay a TCP throughput optimization methodology that selects link-layer transmission modes (e.g. modulation scheme, coding rate, transmission power, or a combination thereof) in accordance with TCP dynamics and wireless channel conditions. We provide insights into the working of the optimization procedure which protects TCP segments against losses on the wireless channel when the TCP congestion window size (in bytes) is below the bandwidth-delay product of the network. The protection against wireless channel losses is rendered by the link-layer by employing robust modulation and coding schemes, high transmission power, etc. We show that TCP dynamics aware link adaptation measures lead to substantial enhancement of TCP throughput in EGPRS and IEEE 802.11a networks

Decision of Maximum Congestion Window Size for TCP Performance Improvement by Bandwidth and RTT Measurement in Wireless Multi-Hop Networks

In the wireless network, TCP performs poorly because it was originally designed for wired networks and does not take into consideration wireless characteristics such as mobility, high-loss probability, and hidden-terminal problems. In particular, in the wireless multi-hop networks, a large congestion window increases the probability of contention and packet losses, and TCP performance is degraded severely as a result. So, it is necessary to limit the TCP congestion window size in order keep the probability of contention loss in the system to a minimum. In this paper, we propose a new scheme for determining the maximum congestion window size based on the measured bandwidth and Round- Trip-Time (RTT). Using ns-2 simulation, we show that the proposed scheme reduces the probability of packet contention and improves TCP performance.

The origins of network server latency &amp; the myth of connection scheduling

We investigate the origins of server-induced latency to understand how to improve latency optimization techniques. Using the Flash Web server [4], we analyze latency behavior under various loads. Despite latency profiles that suggest standard queuing delays, we find that most latency actually originates from negative interactions between the application and the locking and blocking mechanisms in the kernel. Modifying the server and kernel to avoid these problems yields both qualitative and quantitative changes in the latency profiles -- latency drops by more than an order of magnitude, and the effective service discipline also improves.We find our modifications also mitigate service burstiness in the application, reducing the event queue lengths dramatically and eliminating any benefit from application-level connection scheduling. We identify one remaining source of unfairness, related to competition in the networking stack. We show that adjusting the TCP congestion window size addresses this problem, reducing latency by an additional factor of three.

액티브 라우터의 피드백 메커니즘을 이용한 혼잡제어 기법

기존의 end-to-end 방식에서는 네트워크 내부에서 혼잡(congestion)이 발생했을 경우 각 전송자가 즉시 알아 낼 수 없기 때문에 일정시간 동안 수신된 패킷(packet)의 순서에 대한 정보로 흔잡이 발생했는지에 대해 추론하는 것이다. 이와 같은 방법은 RTT(Round Trip Time)가 커지면 혼잡이 발생할 경우 전송자가 전송 양을 줄인다 해도 이미 전송된 패킷들로 인하여 흔잡이 가중되며 전체적인 TCP 동기화 (TCP Global synchronization) 현상을 피할 수 없게 된다. 반면 네트워크 내부에서 직접적으로 정보를 얻거나 처리를 해 줄 수 있다면 혼잡 발생과 동시에 처리가 가능함으로 기존 방식보다 처리율이 향상될 것이다. 본 논문에서는 액티브 라우터의(Active Network) 피드백 메커니즘을 이용하여 네트워크 내부 정보를 각 전송자가 이용할 수 있도록 하기 위해 라우터와의 통신을 이용하였으며, 코어 라우터의 큐 모듈은 RED(Random Early Detection)를 응용하여 ACC의 누락 메커니즘을 개선하였다. ACC를 확장한 메커니즘인 EACC(Enhanced Active Congestion Control)를 제시하고 모의실험을 통해 기존의 혼잡제어나 ACC(Active Congestion Control)보다 성능이 향상됨을 보여준다. Current end-to-end congestion control depends only on the information of end points (using three duplicate ACK packets) and generally responds slowly to the network congestion. This mechanism can't avoid TCP global synchronization which TCP congestion window size is fluctuated during congestion occurred and if RTT (Round Trip Time) is increased, three duplicate ACK packets is not a correct congestion signal because congestion maybe already disappeared and the host may send more packets until receive the three duplicate ACK packets. Recently there is increasing interest in solving end-to-end congestion control using active network frameworks to improve the performance of TCP protocols. ACC (Active congestion control) is a variation of TCP-based congestion control with queue management In addition traffic modifications nay begin at the congested router (active router) so that ACC will respond more quickly to congestion than TCP variants. The advantage of this method is that the host uses the information provided by the active routers as well as the end points in order to relieve congestion and improve throughput. In this paper, we model enhanced ACC, provide its algorithm which control the congestion by using information in core networks and communications between active routers, and finally demonstrate enhanced performance by simulation.

TCP window control scheme using ER value in TCPover ABR services

A window control scheme in TCP over ABR service is proposed. In this scheme, the TCP congestion window size is determined by explicit rate (ER) values, representing the degree of network congestion. Simulation results show an improvement in TCP throughput, reducing the transmissions as well as effectively utilising the bandwidth available to a given TCP connection.