Simulation-based Comparisons of HighSpeed, Westwood and FAST TCP

Size: px
Start display at page:

Download "Simulation-based Comparisons of HighSpeed, Westwood and FAST TCP"

Transcription

1

2 Simulation-based Comparisons of HighSpeed, Westwood and FAST TCP Vance Chen Computer Science Department Boston University Boston, MA ABSTRACT The objective of this project is to study and compare the performances of HighSpeed TCP [1], TCP Westwood [2], and FAST TCP [3] in high speed networks. High speed networks which are envisioned to reliably carry massive amount of data are emerging and becoming more common and widely used (e.g. Gigabit LAN, Internet 2). In a high speed environment we are facing a growing need for more effective communication mechanisms compared to the current in-used TCP (i.e. TCP Reno). A lot of research proposals have discussed the associated evaluation criteria and protocol strategies for the suggested modifications. Solutions are aimed to improve the efficiency of the traditional congestion control mechanism. We used the real time network simulation tool ns2 [4] to complete our analyses of the above three different variants of TCP. First, we were interested in how HighSpeed TCP, TCP Westwood and FAST TCP perform in terms of the throughputs when high packet drop rates presented, the link utilization with effect of different RTTs, their fairness and time to converge and the reaction when only limited buffer available at the router in the high speed network environment. At last, based on the experiments that we have performed and the satisfying results that we have obtained, conclusion has been made that FAST TCP provides better solutions than HighSpeed TCP and TCP Westwood do and is more suitable for today s high speed networks. 1. INTRODUCTION Transmission Control Protocol (TCP) is a key transport protocol in today s Internet. The congestion avoidance mechanisms of the current standard TCP constrains the congestion window that can be achieved by TCP in realistic environments. For example, for a standard TCP connection with byte packets and a 100 ms Round-Trip Time (RTT), achieving a steady-state throughput of 10 Gbps would require an average congestion window of 83,333 segments and a packet drop rate of at most one congestion event every 5,000,000,000 packets. This constraint has severe impacts on performance in a high speed scenario. When a loss is detected either through duplicate acknowledgements (DUPACKs), or through a coarse timeout expiration, the connection backs off by shrinking its congestion window. If the loss is indicated by DUPACKs, TCP Reno attempts to perform a fast recovery by retransmitting the lost segments and halving the congestion window. If the loss is followed by a coarse timeout expiration, the congestion window is reset to1. In either case, after the congestion window is reset, the connection needs several round-trip times before the windowbased probing is restored to near-capacity. As a result, TCP Reno has difficulties in networks with large bandwidth-delay products. At the packet level, linear increase by one packet per RTT is too slow, and multiplicative decrease per loss event is too drastic. At the flow level, maintaining large average congestion windows requires an extremely small equilibrium loss probability. Our goal is to investigate the performance of various newly proposed versions of TCP (in particular, HighSpeed TCP with Limited Slow-Start, TCP Westwood and FAST TCP) in networks of high-speed links (Gbps). To reach our goal, we used the ns2 network simulator. We divided our work

3 into four tasks with each one illustrating different subjects we concerned. This paper is organized as follows: In Section 2 we introduce Limited Slow-Start [5] and HighSpeed TCP which deal with the traditional slow-start and congestion window limitations in high-speed networks. Section 3 and 4 describe the basics of TCP Westwood and FAST TCP. In Section 5 we discuss the related works of other flavors of TCP in high speed environments. Section 6 is our simulation results and we have our conclusion in Section LIMITED SLOW-START and HIGHSPEED TCP Slow-Start is a mechanism to avoid that an inappropriate large burst of traffic is suddenly put into the pipe when the sender begins to send its data through a newly-created TCP connection. The basic idea is that the sending rate at which new packets should be injected into the network depends on the rate at which the acknowledgements (ACKs) are returned by the receiver. The original Slow-Start works as follow. At the beginning, the parameter cwnd is set to one segment size and the parameter ssthresh is set to receiver s advertised window value. The sender can send up to the minimum of cwnd and the receiver s advertised window. It transmits one segment and waits for the corresponding ACK. When the ACK is received, the congestion window is incremented from one to two, and two new segments can be sent. When each of those two segments is acknowledged, the congestion window is incremented to four and so on resulting in an exponential growth of cwnd. Slow-Start continues until cwnd meets or exceeds the ssthresh or a packet loss is detected. The problem with original Slow-Start is that in high speed network environments, cwnd must reach values of hundreds even thousands of segments and then the current Slow-Start can lead to increase the congestion window by thousands of segments per RTT. Such an increase can easily overwhelm the available bandwidth and result in thousands of packets to be dropped in just one RTT which need to be recovered later on. Limited Slow-Start is a modified version of traditional Slow-Start proposed by Sally Floyd trying to provide a solution to the above problem. Limited Slow-Start introduces a parameter, max_ssthresh, and modifies the slow-start mechanism for values of the congestion window. For each arriving ACK in Slow-Start phase: If (cwnd <= max_ssthresh) cwnd += MSS; // Standard Slow-Start else K = int(cwnd/(0.5 max_ssthresh)); cwnd += int(mss/k); // Limited Slow-Start Namely, Limited Slow-Start equals Standard Slow- Start if cwnd <= max_ssthresh and cwnd is increased by one Maximum Segment Size (MSS) for every arriving ACK. Limited Slow-Start takes effect if max_ssthresh < cwnd <= ssthresh and then cwnd is increased by at most max_ssthresh/2 MSS per round trip time. Thus, during Limited Slow- Start the window is increased by 1/K MSS for each arriving ACK, for K = int(cwnd/(0.5 max_ssthresh)), instead of by 1 MSS as in standard slow-start. When ssthresh < cwnd, the TCP sender exits Slow-Start, and enters the congestion avoidance phase. The author of Limited Slow-Start recommended setting max_ssthresh to be 100 MSS. The traditional congestion avoidance algorithm uses Additive Increase Multiplicative Decrease (AIMD) to probe the available bandwidth and the network for additional capacity gradually. During this phase, cwnd is increased one segment size per RTT and halved if a packet loss is detected. In Section 1, we have already described the problems of congestion avoidance in TCP Reno in high speed networks. HighSpeed TCP is a sender-side alteration to regular TCP. It was introduced by Sally Floyd as a modification of TCP s congestion control mechanism to improve the performance in fast, long delay networks. This modification is designed to behave like Reno for small values of cwnd, but above a chosen value of cwnd a more aggressive response function is used (Table 1 and Table 2). When cwnd is large (greater than 38 packets), the modification uses a table to indicate by how much to increase the congestion window when an ACK is received, and it releases less network bandwidth than 1/2 cwnd on packet losses. The different response function achieves high throughput without requiring unrealistically low packet drop rates. HighSpeed TCP increases the congestion window by a(w) segments per RTT in the absence of congestion, and decreases the congestion window to w(1 b(w)) segments in response to a RTT with

4 one or more loss events. Thus in response to a single ACK, HighSpeed TCP increases its cwnd in segments as follows: w w + a(w)/w and responds to a congestion event by decreasing the congestion window as w w(1 b(w)). In standard TCP, a(w) = 1 and b(w) = 1/2. Basically, a(w) is a function of the current window size and the packet drop rate and b(w) is a function of the current window size. Packet Drop Congestion RTTs between Rate P Window W Losses 10 ^ ^ ^ ^ ^ ^ ^ ^ ^ Table 1: TCP Response Function for Standard TCP (TCP Reno). The average congestion window W in MSS-sized segments is given as a function of the packet drop rate P. Packet Drop Congestion RTTs between Rate P Window W Losses 10 ^ ^ ^ ^ ^ ^ ^ ^ ^ Table 2: TCP Response Function for HighSpeed TCP. The average congestion window W in MSS-sized segments is given as a function of the packet drop rate P. In Figure 1 we showed both the values of cwnd for HighSpeed TCP and regular TCP. We can see in HighSpeed TCP the more aggressive AIMD mechanism is applied to achieve the high throughput more effectively compared to the regular TCP. Overall, HighSpeed TCP s goal is to allow TCP to achieve high throughput with more realistic requirements for the steady-state packet drop rate. HighSpeed TCP has more realistic requirements for the number of RTTs between loss events [1]. Figure 1: cwnd of HighSpeed TCP and regular TCP 3. TCP WESTWOOD TCP Westwood is a new protocol that modifies the sender-side of the window congestion control scheme. The strategy is that TCP Westwood sets a slow start threshold and a congestion window which are consistent with the network capacity measured at the time congestion is experienced. It controls the window by using end-to-end connection bandwidth share estimation. TCP Westwood takes continuous estimates, at the TCP sender, of the packet rate of the connection. This is done by monitoring the ACK reception rate. DUPACKs, indicate an out-of-order reception and they should count toward the bandwidth estimate, and new estimate should be computed right after their reception. However, the sender is in no position to tell for sure which segment triggered the DUPACK transmission, and it is thus unable to update the data count by the size of that segment. The authors of TCP Westwood first assumed all TCP segments have the same size and the sequence numbers are incremented by one per segment sent and then used an average of the segment size sent thus far in the ongoing connection which allows for corrections when the next cumulative ACK is received. The basic bandwidth estimation is performed using a low-pass filter described by the following pseudo code:

5 If (ACK is received) sample_bwe k = (acked * pkt_size * 8) / (now - lastacktime); BWE k = (19 / 21) * BWE k-1 + (1 / 21) * (sample_bwe k + sample_bwe k - 1 ); End if In the algorithm, acked is the number of segments acknowledged by the latest ACK. The authors also took into account the effects of delayed and cumulative ACKs and took care of the issue of DUPACKs mentioned earlier for a more accurate acked variable in the above algorithm by using the procedure AckedCount as described in [2]. pkt_size is the segment size in bytes, now indicates the current time and lastacktime indicates the time that the previous ACK was received. Therefore, BWE is the low-pass filtered measurement of the available bandwidth. The BWE as defined above is the rate actually achieved by the individual flow. When the bottleneck becomes saturated and packets are dropped, TCP Westwood selects congestion windows which enforce the current measured rates. In particular, assuming that a sender has determined the connection bandwidth estimate, it is used to properly set cwnd and ssthresh after a packet loss indication. In TCPWestwood, the congestion window dynamics during slow start and congestion avoidance are unchanged, which means that they increase exponentially and linearly, respectively, as in current TCP Reno. The idea of TCP Westwood is to try to compute each flow s fair share of the bandwidth directly, rather than performing the traditional slow start and congestion control periods. This will improve performance in high speed networks because it will allow the TCP flows to converge to their share of the bandwidth at a much faster rate. In Figure 2 we showed the link utilization of TCP WestwoodNR (the name followed by NR because it is based on the TCP New Reno) and TCP New Reno. As we can see TCP WestwoodNR reached the full capacity of the link much faster than TCP New Reno. TCP WestwoodNR used less than 1 second to get the full bandwidth. However, TCP New Reno used about 10 seconds to reach the full capacity. Figure 2: Link utilization of TCP WestwoodNR and TCP New Reno 4. FAST TCP FAST TCP can be viewed as a high speed version of TCP Vegas. It was introduced by Steven Low and his group at Caltech. The window control component determines the congestion window based on both queuing delays and packet losses. The key decision of the design is that the same algorithm is used for congestion window computation independent of the state of the sender. FAST TCP is improved based on a prime-dual model where TCP protocol is modeled by a nonlinear closed-feedback and time-delay control system. Current TCP is not stable when used in a network with high product of capacity and delay. Therefore, to stabilize the sending rate at the source, FAST TCP implements two control mechanisms. One is implemented at the sender to adjust the sending rate dynamically based on an equation and another one is to obtain a congestion measure based on the aggregate flow rate on a link. The equation for adjusting sending rate at the sender is obtained by proper parameter assignment and polezero placement using Nyquist stability analysis. By appropriately choosing the equation, FAST TCP can achieve its objective for high performance, stability and fairness in high speed environments. Under normal network conditions, FAST TCP periodically updates the congestion window as followed: cwnd min{2 * cwnd, (1 - gamma) * cwnd + gamma * (basertt / RTT + alpha (cwnd, qdelay))} based on the average RTT and average queuing delay provided by the estimation

6 component, where gamma = (0, 1], basertt is the minimum RTT observed so far, RTT is the sample RTT measured by a weighted averaging formula, qdelay is the end-to-end (average) queuing delay, and the constant alpha is the number of packets each flow attempts to maintain in the network buffers at equilibrium, similar to TCP Vegas. Currently when a packet loss is detected, FAST TCP halves its window in order to back off packet transmission quickly when sever congestion occurs and bring the system back to the regime where reliable RTT measurements are again available for window adjustment (using the formula described above) to work effectively. Details on how the estimations for average RTT and queuing delay are done could be found in [3]. In Figure 3, we showed the comparison of link utilization for FAST TCP and TCP Vegas. As we can see FAST TCP reached the full capacity of the link much faster than TCP Vegas. FAST TCP used only less than 1 second to reach the full capacity of the link. However, TCP Vegas needed around 10 seconds to get the full bandwidth. recovery time of the standard TCP. The idea is built on the idea of HighSpeed TCP. Packet loss recovery times for a traditional TCP connection as well as HighSpeed TCP connection are proportional to the connection s congestion window size and RTT. A Scalable TCP connection s packet loss recovery times are proportional to connection s RTT only. The generalized TCP window update algorithm responds to each acknowledgement received in a round trip time in which congestion has not been detected with the update by cwnd cwnd + a where a is a constant between 0 and 1. When the first detection of congestion is given in a round trip time, the congestion window is altered by cwnd cwnd ceil(b * cwnd) where b is also a constant between 0 and 1. The values of a and b are selected by considering the convergence speed and instantaneous rate variation and can be configured using the proc file system by a super user in Linux. Like HighSpeed TCP the modified congestion avoidance algorithm takes place only when the congestion window is above some threshold. The default values for a is 0.01, b is 0.125, and the threshold congestion window size is 16 segments. Overall Scalable TCP was designed from a strong theoretical base in order to ensure resource sharing and stability while also remaining agile in prevailing network conditions. Figure 4 (from Scalable TCP website shows the main difference in the scaling properties of traditional and Scalable TCP. Figure 3: Link utilization of FAST TCP and TCP Vegas 5. RELATED WORK In this section, we describe two additional TCP variants designed for high speed network environments. 5.1 Scalable TCP Scalable TCP [6] is a sender-side alteration to the regular TCP congestion window update algorithm. The main goal of Scalable TCP is to improve the loss Figure 4a: Traditional TCP with large capacity

7 6. SIMULATION RESULTS Our objective for this project is to use network simulator ns-2 in order to thoroughly understand the issues that take place over high speed links. We have designed a series of cases that will help us understand every detail of what takes place in a network with high speed links when using HighSpeed TCP with Limited Slow-Start, TCP WestwoodNR, and FAST TCP. 6.1 Packet Drop Rate Figure 4b: Scalable TCP with large capacity 5.2 XCP (explicit Congestion control Protocol) XCP [7] was proposed by Dina Katabi at el. in MIT. XCP generalized the Explicit Congestion Notification (ECN). Instead of one bit congestion indication used by ECN, it proposes using precise congestion signaling, where the network explicitly tells the sender the state of congestion and how to react to it. Sender maintains the congestion window and RTT and communicates this to routers via a congestion header (Figure 5) in every packet. Then it uses the feedback field in the congestion header to request its desired window increase. In addition, XCP introduces the new concept of decoupling utilization control from fairness control. A router has both an efficiency controller and fairness controller. This allows a more flexible and analytically tractable protocol design and opens new avenues for service differentiation. The purpose of efficiency controller is to maximize link utilization while minimizing drop rate and persistent queues. The fairness controller uses standard TCP s AIMD mechanism to converge to fairness. Sender s current cwnd (filled by sender and remains unmodified) Sender s RTT estimate (filled by sender and remains unmodified) Feedback (initialized to sender s demands; can be modified by the routers) Figure 5: Congestion header in a packet for XCP There are several reasons can cause high packet drop rate. Even in high speed networks, when too many flows are active and the bandwidth or the queue size of the bottleneck link is limited, TCP should be responsible for dealing with this issue properly to maintain the throughput and achieve the stability. We tried to analyze the performance of HighSpeed TCP, TCP WestwoodNR, and FAST TCP with high sending rate in the presence of high packet drop rate. The related issue, buffer management will be discussed in Section 6.4. Network Topology: S 1 Gbps 2 ms delay r Mbps 10 ms delay r 2 1 Gbps 2 ms delay In this section we simulated HighSpeed TCP, TCP WestwoodNR and FAST TCP to see how they perform under different packet drop rates in high speed networks. Here in this case we set the buffer size to a fixed size (bandwidth-delay product) to fully utilize the bottleneck link. Packets would be dropped due to buffer overflow. In addition, we used an instance of ErrorModel in ns-2 to artificially generate packet losses. The instance we created was a random variable with uniform distribution. We were particularly interested in and specified the packet drop rates as 10 ^ -7, 10 ^ -5 and 10 ^ -3. The receiver s window is set to a large enough size so that it will not limit the bandwidth usage on the link. Ideally, the modified congestion control mechanisms of these newly-proposed versions of TCP should properly response to different packet loss rates in the network of high speed links. R

8 Figure 5a: Link utilization of HighSpeed TCP, TCP Westwood and FAST TCP with packet drop rate = 10 ^ -7 Figure 5b: Link utilization of HighSpeed TCP, TCP Westwood and FAST TCP with packet drop rate = 10 ^ -5 Figure 5c: Link utilization of HighSpeed TCP, TCP Westwood and FAST TCP with packet drop rate = 10 ^ -3

9 HighSpeed TCP: Network Topology: In Figure 5a and Figure 5b, we can see that HighSpeed TCP performed very well when the loss rate ranges from very low (10 ^ -7) to medium (10 ^ - 5). However, when the loss rate is high (10 ^ -3 in Figure 5c), HighSpeed TCP only can utilize about 30% of the bandwidth of the bottleneck link. Insuch a high loss rate environment, HighSpeed TCP performed worse than both TCP WestwoodNR and FAST TCP. S1 S2 1 Gbps 2 ms r1 1 Gbps 10 ms 100 Mbps 10 ms 1 Gbps 2 ms r2 1 Gbps 2 ms R1 R2 TCP WestwoodNR: As we can see among the three graphs of Figure 5, even if TCP WestwoodNR did not fully utilize the bottleneck link when setting the loss rate at 10 ^ -7 and 10 ^ -5, due to its better congestion control mechanism than HighSpeed TCP, it still can use at least 50% of the bandwidth of the bottleneck link when the loss rate is very high (10 ^ -3). In comparison to HighSpeed TCP, it is more stable because it is designed not only for high speed networks but also for high packet drop rate environments, such as wireless networks. FAST TCP: It is apparent that FAST TCP performed the best in our packet drop rate experiments. Even if the loss rate is as high as 10 ^ -3, from Figure 5c, we can see that FAST TCP maintained an almost 100% utilization of the bandwidth of the bottleneck link. Beyond that, it also performed the best in the other two cases of different packet drop rate due to its better queue management and RTT estimation based on TCP Vegas. 6.2 Effect of RTT The effect of RTT is an interesting subject especially when FAST TCP which is based on TCP Vegas uses it to estimate the available bandwidth and adjust its sending rate. We would like to see the stability of HighSpeed TCP, TCP WestwoodNR, and FAST TCP in the presence of different RTT in high speed networks. In this section we simulated HighSpeed TCP, TCP WestwoodNR and FAST TCP and analyzed the effects of changing the delays on each of the incoming links to the bottleneck link to see how the performance would be affected. We set the RTT of flow 1 from S1 to R1 through router r1 and r2 28 ms and set the RTT of flow 2 from S2 to R2 also through router r1 and r2 44 ms. In this experiment, flow 1 and flow 2 belonged to the same version of TCP and packet will be dropped due to buffer overflow at the bottleneck link. The total simulation time is 100 seconds HighSpeed TCP: From Figure 6a, flow 1 with 28 ms RTT dominated the 90% of the bandwidth of the bottleneck link through the whole simulation process. Flow 2 with longer RTT (44 ms) only got around 10% of the bandwidth of the bottleneck link. Even if it is very stable for those two HighSpeed TCP flows (their allocated bandwidth almost did not change and the cwnd of each flow also fixed in certain range), however we did not consider the two flows got fair share of the bandwidth allocation. The term fair share and fairness we used in this paper mean max-min fairness. That is we considered the modified version of TCP satisfied fairness if and only if flow 1 and flow 2 obtained equal bandwidth (50% for each flow) at the bottleneck link because we only had two flows competing here and they were set to have the same sending rates. However for HighSpeed TCP, because the connection with shorter RTT reacts sooner, it would get more bandwidth. This is also true for TCP WestwoodNR as we would see. It is the problem of RTT-proportional fairness presented in traditional TCP as well.

10 Figure 6a: Link utilization of two HighSpeed TCP flows with different RTTs Figure 6b: Link utilization of two TCP Westwood flows with different RTTs Figure 6c: Link utilization of two FAST TCP flows with different RTTs

11 TCP WestwoodNR: Figure 6b showed that the dramatic oscillations for both TCP WestwoodNR flows. Not even did the two flows with different RTTs get their fair share of the bandwidth, but very unstable in the value of cwnd. At the beginning it seemed that the two flows tended to get the equal share of the bandwidth allocation of the bottleneck link. However, TCP WestwoodNR failed to stabilize the two flows with their different RTTs during the whole simulation process. We consider TCP WestwoodNR the worst in this test. FAST TCP: Again, FAST TCP which based on TCP Vegas performed superiorly in the RTT test case. In Figure 6c, we can clearly see the two FAST TCP connections with different RTTs resulted in an exactly equal 50% bandwidth allocation of the bottleneck link for each flow at the beginning till the end of the simulation. Worth to note, the convergence only took about 1 second. Then we got very stable link utilization and cwnd for both flows. The performance of FAST TCP is not affected by different RTTs. As the result, we believe FAST TCP is better when handling flows with different RTT 6.3 Fairness and Convergence Time The main problem with high speed links is that because of the regular AIMD congestion control mechanism it may take some flows a long time to adapt to their share of the bandwidth. One question in high speed networks is the convergence time for the TCP flows and when these flows converge do they get their fair share of the bandwidth. The objective of this experiment is to calculate how long it takes for a flow to use all of the available bandwidth. The tests are when one flow starts before the other, how long it takes for the two flows to get to their fair share of the bandwidth and then when one of the flows ends before the other how long it will take for the remaining flow to get to the point where it is using all of the available bandwidth. Network Topology: The same as Section 6.2, we only change the RTT from S2 to r1 to 2 ms to get the equal RTTs for both flows. In this section, we performed several experiments and each of them took 100 seconds. First we started two HighSpeed TCP at the same time and ended one of them at 60 th second to see how fast the other flow can utilize the full capacity of the bandwidth. Second, we started only one flow at the beginning and after 20 seconds we added the second flow to see how fast the flows converge. Then we did the same tests on TCP WestwoodNR and FAST TCP. HighSpeed TCP: When one of the same starting time flows ended at 60 th second, the other flow can immediately utilize the full bandwidth of the bottleneck link. That means HighSpeed TCP responses to the change of the status of a high speed network very quickly when there is only one flow left. However, when there was only one flow in the high speed network and we added the other at 20 th second, HighSpeed TCP performed the worst when trying to converge. The two flows took over 55 seconds to converge. That is, we started the second flow at 20 th second but the first flow kept using most of the bandwidth till the two flows converged at nearly 80 th second. Compared to TCP WestwoodNR and FAST TCP, HighSpeed TCP took too much time for convergence. From Figure 7a we can see also the number of oscillation is really high. We had a congestion epoch about every 1.5 second for each flow. Even if HighSpeed TCP roughly maintained a fair share for each connection, the bandwidth allocated to each connection was varying from 30% to 70%. TCP WestwoodNR: For TCP WestwoodNR, when one of the same starting time flows ended at 60 th second, the other flow can immediately utilize the full bandwidth of the bottleneck link though with a little bit delay compared to HighSpeed TCP and FAST TCP. That means TCP WestwoodNR responses to the status of the high speed environments quickly. Even when there was only one flow in the high speed network and we added the other at 20 th second, TCP WestwoodNR performed very well when trying to converge. The two flows took less than 20 seconds to converge. That is, we started the second flow at 20 th second and the two flows converged at around 38 th second. Compared to HighSpeed TCP, TCP WestwoodNR took much less time for convergence. From Figure 7b we can see not only the convergence time for two TCP WestwoodNR connections was short, but the number of oscillation is really less than it in HighSpeed TCP as well. We had a congestion epoch about every 5~6 second for each flow. As to fairness, TCP WestwoodNR had similar fairness for each connection as HighSpeed TCP did. It roughly maintained a fair share for each flow, but with the bandwidth allocated to each connection was varying from 30% to 70%.

12 Figure 7a: Link utilization of two HighSpeed TCP flows with different ending time Link utilization of two HighSpeed TCP flows with different starting time Figure 7b: Link utilization of two TCP WestwoodNR flows with different ending time Link utilization of two TCP WestwoodNR flows with different starting time Figure 7c: Link utilization of two FAST TCP flows with different ending time Link utilization of two FAST TCP flows with different starting time

13 FAST TCP: The simulation results here for FAST TCP are very satisfying. FAST TCP allocated equally fair share (50%) to both flows at the beginning and when one of the two flows ended at 60 th second, the other flow reached the full bandwidth of the bottleneck link immediately. Similarly, when we started one flow first and then added the other flow at 20 th, those two flows converged immediately and got their fair share about 50% of the available bandwidth. In Figure 7c we have shown that when considering the fairness and the convergence time in a high speed environment. FAST TCP performed superiorly than HighSpeed TCP and TCP WestwoodNR. 6.4 Buffer Management Buffer overflow is the main reason to cause the packets to be dropped at the bottleneck link. Normally, the buffer size is set to be the bandwidth delay product of the bottleneck link. However, we would like to see how the different versions of TCP react to a limited buffer size in the router at the bottleneck link in the high speed networks. We are also interested in the effect of doubling the bandwidth delay product buffer size in the network of high speed links. Network Topology: The same as Section 6.3, we only change the queue size at the bottleneck link from r1 to r2. The goal of this experiment is to see how the queue size affects the performance of each of these three versions of TCP and namely how those TCP manage the buffer in the high speed environments. First we set the buffer size as half of the bandwidth delay product, and then we set the buffer size as double the bandwidth delay product. The same simulations were performed on HighSpeed TCP, TCP WestwoodNR, and FAST TCP. HighSpeed TCP: When the buffer size was half of the bandwidth delay product, the two HighSpeed TCP flows can reach totally only 95% bandwidth of the bottleneck link. Each of the two connections utilized the bandwidth from 30% to 65%. When the buffer was doubled, even 100% utilization of the capacity of the link could be reached, each flow got its share varied dramatically from 0 to 1. HighSpeed TCP performed the worst in this test case. TCP WestwoodNR: Figure 8b showed that halving the buffer size, TCP WestwoodNR still can perform very well. Also more stable than HighSpeed TCP no matter it is the case we halved the buffer size or doubled the buffer size. The utilization of the bottleneck link could be maintained at 100% and the fairness was good enough to let each flow get around 50% of the bandwidth. However, when the queue size was doubled, each of the WestwoodNR connections may get its share ranging from 30% to 70%. Generally, TCP WestwoodNR has better queue management than HighSpeed TCP due to its continuously bandwidth estimation. FAST TCP: Again, we can see from Figure 8c, FAST TCP has the best queue management scheme. Each of the two flows got almost exactly 50% of the bandwidth all the time regardless of the queue size at the bottleneck link. It is also shown that FAST TCP is very stable while we changing the size of the buffer. FAST TCP like TCP Vegas can usually maintain certain number of packets kept in the queue in some fixed range. Due to so, FAST TCP manages the queue more effectively than HighSpeed TCP and TCP Westwood and thus performed the best in this test case. 7. CONCLUSION In this project, we introduced some lately proposal for improving TCP performance in high speed networks. Especially, we described the basic ideas of HighSpeed TCP with Limited Slow-Start, TCP Westwood, and FAST TCP. Then we designed some experiments we were interested in. Each of the tasks focuses on a different aspect of performance in a network with high speed links. In 6.1 we evaluated the performance of each of the three versions of TCP individually with different packet drop rate. We took into account the effect of RTTs in 6.2. Then we took care of the convergence time and the fairness of HighSpeed TCP, TCP Westwood, and FAST TCP in the high speed environments. At last, we performed the simulation on these TCP with different buffer size. All the results we obtained from the simulation have shown that FAST TCP outperformed the other two in the test cases we set. Its performance is not affected by the RTT and can reach high throughput even under high packet drop rate. FAST TCP flows also converge to their fair share in a very short of time and can manage any size of the queue at the bottleneck link effectively.

14 Figure 8a: Link utilization of two HighSpeed TCP flows with half bandwidth delay product queue size Link utilization of two HighSpeed TCP flows with double bandwidth delay product queue size Figure 8b: Link utilization of two TCP WestwoodNR flows with half bandwidth delay product queue size Link utilization of two TCP WestwoodNR flows with double bandwidth delay product queue size Figure 8c: Link utilization of two FAST TCP flows with half bandwidth delay product queue size Link utilization of two FAST TCP flows with double bandwidth delay product queue size

15 8. REFERENCES [1] Sally Floyd, HighSpeed TCP for Large Congestion Windows, RFC 3649, Experimental, December 2003 [2] S. Mascolo, C. Casetti, M. Gerla, S. S. Lee, and M. Sanadidi, TCP Westwood: congestion control with faster recovery, UCLA CSD Technical Report #200017, 2000 [3] Cheng Jin, David X. Wei and Steven H. Low, FAST TCP: Motivation, Architecture, Algorithms, Performance, Caltech CS Report CaltechCSTR:2003:010, December 17, 2003 [4] The Network Simulator ns-2 [5] Sally Floyd, Limited Slow-Start for TCP with Large Congestion Windows, RFC 3742, Experimental, March 2004 [6] Tom Kelly, Scalable TCP: Improving Performance in High Speed Wide Area Networks, First International Workshop on Protocols for Fast Long-Distance Networks, Geneva, February 2003 [7] Dina Katabi, Mark Handley and Chalrie Rohrs, Congestion Control for High Bandwidth-Delay Product Networks, Proceedings on ACM Sigcomm 2002 [8] M. Gerla, M. Y. Sanadidi, R. Wang, A. Zanella, C. Casetti, S. Mascolo, "TCP Westwood: Congestion Window Control Using Bandwidth Estimation", In Proceedings of IEEE Globecom 2001, Volume: 3, pp , San Antonio, Texas, USA, November 25-29, 2001 [9] M. Gerla, S. S. Lee, and G. Pau, "TCP Westwood Simulation Studies in Multiple-Path Cases", Proc. International Symposium on Performance Evaluation of Computer and Telecommunication Systems (SPECTS), San Diego, CA, USA, July, 2002 [10] NS2 implementation of FAST TCP [11] Cheng Jin, David X. Wei and Steven H. Low, FAST TCP for high-speed long-distance networks, Internet-Draft, June 30, 2003

Improving the Performance of TCP Using Window Adjustment Procedure and Bandwidth Estimation

Improving the Performance of TCP Using Window Adjustment Procedure and Bandwidth Estimation Improving the Performance of TCP Using Window Adjustment Procedure and Bandwidth Estimation R.Navaneethakrishnan Assistant Professor (SG) Bharathiyar College of Engineering and Technology, Karaikal, India.

More information

A Survey on Congestion Control Mechanisms for Performance Improvement of TCP

A Survey on Congestion Control Mechanisms for Performance Improvement of TCP A Survey on Congestion Control Mechanisms for Performance Improvement of TCP Shital N. Karande Department of Computer Science Engineering, VIT, Pune, Maharashtra, India Sanjesh S. Pawale Department of

More information

Data Networks Summer 2007 Homework #3

Data Networks Summer 2007 Homework #3 Data Networks Summer Homework # Assigned June 8, Due June in class Name: Email: Student ID: Problem Total Points Problem ( points) Host A is transferring a file of size L to host B using a TCP connection.

More information

TCP over Multi-hop Wireless Networks * Overview of Transmission Control Protocol / Internet Protocol (TCP/IP) Internet Protocol (IP)

TCP over Multi-hop Wireless Networks * Overview of Transmission Control Protocol / Internet Protocol (TCP/IP) Internet Protocol (IP) TCP over Multi-hop Wireless Networks * Overview of Transmission Control Protocol / Internet Protocol (TCP/IP) *Slides adapted from a talk given by Nitin Vaidya. Wireless Computing and Network Systems Page

More information

TCP in Wireless Mobile Networks

TCP in Wireless Mobile Networks TCP in Wireless Mobile Networks 1 Outline Introduction to transport layer Introduction to TCP (Internet) congestion control Congestion control in wireless networks 2 Transport Layer v.s. Network Layer

More information

Lecture 15: Congestion Control. CSE 123: Computer Networks Stefan Savage

Lecture 15: Congestion Control. CSE 123: Computer Networks Stefan Savage Lecture 15: Congestion Control CSE 123: Computer Networks Stefan Savage Overview Yesterday: TCP & UDP overview Connection setup Flow control: resource exhaustion at end node Today: Congestion control Resource

More information

An Improved TCP Congestion Control Algorithm for Wireless Networks

An Improved TCP Congestion Control Algorithm for Wireless Networks An Improved TCP Congestion Control Algorithm for Wireless Networks Ahmed Khurshid Department of Computer Science University of Illinois at Urbana-Champaign Illinois, USA khurshi1@illinois.edu Md. Humayun

More information

Low-rate TCP-targeted Denial of Service Attack Defense

Low-rate TCP-targeted Denial of Service Attack Defense Low-rate TCP-targeted Denial of Service Attack Defense Johnny Tsao Petros Efstathopoulos University of California, Los Angeles, Computer Science Department Los Angeles, CA E-mail: {johnny5t, pefstath}@cs.ucla.edu

More information

Chaoyang University of Technology, Taiwan, ROC. {changb,s9227623}@mail.cyut.edu.tw 2 Department of Computer Science and Information Engineering

Chaoyang University of Technology, Taiwan, ROC. {changb,s9227623}@mail.cyut.edu.tw 2 Department of Computer Science and Information Engineering TCP-Taichung: A RTT-based Predictive Bandwidth Based with Optimal Shrink Factor for TCP Congestion Control in Heterogeneous Wired and Wireless Networks Ben-Jye Chang 1, Shu-Yu Lin 1, and Ying-Hsin Liang

More information

Lecture Objectives. Lecture 07 Mobile Networks: TCP in Wireless Networks. Agenda. TCP Flow Control. Flow Control Can Limit Throughput (1)

Lecture Objectives. Lecture 07 Mobile Networks: TCP in Wireless Networks. Agenda. TCP Flow Control. Flow Control Can Limit Throughput (1) Lecture Objectives Wireless and Mobile Systems Design Lecture 07 Mobile Networks: TCP in Wireless Networks Describe TCP s flow control mechanism Describe operation of TCP Reno and TCP Vegas, including

More information

Chapter 6 Congestion Control and Resource Allocation

Chapter 6 Congestion Control and Resource Allocation Chapter 6 Congestion Control and Resource Allocation 6.3 TCP Congestion Control Additive Increase/Multiplicative Decrease (AIMD) o Basic idea: repeatedly increase transmission rate until congestion occurs;

More information

CUBIC: A New TCP-Friendly High-Speed TCP Variant

CUBIC: A New TCP-Friendly High-Speed TCP Variant : A New TCP-Friendly High-Speed TCP Variant Injong Rhee, and Lisong Xu Abstract This paper presents a new TCP variant, called, for high-speed network environments. is an enhanced version of : it simplifies

More information

Research of TCP ssthresh Dynamical Adjustment Algorithm Based on Available Bandwidth in Mixed Networks

Research of TCP ssthresh Dynamical Adjustment Algorithm Based on Available Bandwidth in Mixed Networks Research of TCP ssthresh Dynamical Adjustment Algorithm Based on Available Bandwidth in Mixed Networks 1 Wang Zhanjie, 2 Zhang Yunyang 1, First Author Department of Computer Science,Dalian University of

More information

Transport Layer Protocols

Transport Layer Protocols Transport Layer Protocols Version. Transport layer performs two main tasks for the application layer by using the network layer. It provides end to end communication between two applications, and implements

More information

Linux 2.4 Implementation of Westwood+ TCP with rate-halving: A Performance Evaluation over the Internet

Linux 2.4 Implementation of Westwood+ TCP with rate-halving: A Performance Evaluation over the Internet Linux. Implementation of TCP with rate-halving: A Performance Evaluation over the Internet A. Dell Aera, L. A. Grieco, S. Mascolo Dipartimento di Elettrotecnica ed Elettronica Politecnico di Bari Via Orabona,

More information

SJBIT, Bangalore, KARNATAKA

SJBIT, Bangalore, KARNATAKA A Comparison of the TCP Variants Performance over different Routing Protocols on Mobile Ad Hoc Networks S. R. Biradar 1, Subir Kumar Sarkar 2, Puttamadappa C 3 1 Sikkim Manipal Institute of Technology,

More information

TCP Westwood for Wireless

TCP Westwood for Wireless TCP Westwood for Wireless מבוא רקע טכני בקרת עומס ב- TCP TCP על קשר אלחוטי שיפור תפוקה עם פרוטוקול TCP Westwood סיכום.1.2.3.4.5 Seminar in Computer Networks and Distributed Systems Hadassah College Spring

More information

Internet Flow Control - Improving on TCP

Internet Flow Control - Improving on TCP Internet Flow Control - Improving on TCP Glynn Rogers The Team - Jonathan Chan, Fariza Sabrina, and Darwin Agahari CSIRO ICT Centre Why Bother? - Isn t TCP About as Good as It Gets?! Well, TCP is a very

More information

XCP-i : explicit Control Protocol for heterogeneous inter-networking of high-speed networks

XCP-i : explicit Control Protocol for heterogeneous inter-networking of high-speed networks : explicit Control Protocol for heterogeneous inter-networking of high-speed networks D. M. Lopez-Pacheco INRIA RESO/LIP, France Email: dmlopezp@ens-lyon.fr C. Pham, Member, IEEE LIUPPA, University of

More information

Application Level Congestion Control Enhancements in High BDP Networks. Anupama Sundaresan

Application Level Congestion Control Enhancements in High BDP Networks. Anupama Sundaresan Application Level Congestion Control Enhancements in High BDP Networks Anupama Sundaresan Organization Introduction Motivation Implementation Experiments and Results Conclusions 2 Developing a Grid service

More information

International Journal of Scientific & Engineering Research, Volume 6, Issue 7, July-2015 1169 ISSN 2229-5518

International Journal of Scientific & Engineering Research, Volume 6, Issue 7, July-2015 1169 ISSN 2229-5518 International Journal of Scientific & Engineering Research, Volume 6, Issue 7, July-2015 1169 Comparison of TCP I-Vegas with TCP Vegas in Wired-cum-Wireless Network Nitin Jain & Dr. Neelam Srivastava Abstract

More information

Optimization of Communication Systems Lecture 6: Internet TCP Congestion Control

Optimization of Communication Systems Lecture 6: Internet TCP Congestion Control Optimization of Communication Systems Lecture 6: Internet TCP Congestion Control Professor M. Chiang Electrical Engineering Department, Princeton University ELE539A February 21, 2007 Lecture Outline TCP

More information

4 High-speed Transmission and Interoperability

4 High-speed Transmission and Interoperability 4 High-speed Transmission and Interoperability Technology 4-1 Transport Protocols for Fast Long-Distance Networks: Comparison of Their Performances in JGN KUMAZOE Kazumi, KOUYAMA Katsushi, HORI Yoshiaki,

More information

High-Speed TCP Performance Characterization under Various Operating Systems

High-Speed TCP Performance Characterization under Various Operating Systems High-Speed TCP Performance Characterization under Various Operating Systems Y. Iwanaga, K. Kumazoe, D. Cavendish, M.Tsuru and Y. Oie Kyushu Institute of Technology 68-4, Kawazu, Iizuka-shi, Fukuoka, 82-852,

More information

APPENDIX 1 USER LEVEL IMPLEMENTATION OF PPATPAN IN LINUX SYSTEM

APPENDIX 1 USER LEVEL IMPLEMENTATION OF PPATPAN IN LINUX SYSTEM 152 APPENDIX 1 USER LEVEL IMPLEMENTATION OF PPATPAN IN LINUX SYSTEM A1.1 INTRODUCTION PPATPAN is implemented in a test bed with five Linux system arranged in a multihop topology. The system is implemented

More information

TCP over Wireless Networks

TCP over Wireless Networks TCP over Wireless Networks Raj Jain Professor of Computer Science and Engineering Washington University in Saint Louis Saint Louis, MO 63130 Audio/Video recordings of this lecture are available at: http://www.cse.wustl.edu/~jain/cse574-10/

More information

Parallel TCP Data Transfers: A Practical Model and its Application

Parallel TCP Data Transfers: A Practical Model and its Application D r a g a n a D a m j a n o v i ć Parallel TCP Data Transfers: A Practical Model and its Application s u b m i t t e d t o the Faculty of Mathematics, Computer Science and Physics, the University of Innsbruck

More information

A Congestion Control Algorithm for Data Center Area Communications

A Congestion Control Algorithm for Data Center Area Communications A Congestion Control Algorithm for Data Center Area Communications Hideyuki Shimonishi, Junichi Higuchi, Takashi Yoshikawa, and Atsushi Iwata System Platforms Research Laboratories, NEC Corporation 1753

More information

A Survey: High Speed TCP Variants in Wireless Networks

A Survey: High Speed TCP Variants in Wireless Networks ISSN: 2321-7782 (Online) Volume 1, Issue 7, December 2013 International Journal of Advance Research in Computer Science and Management Studies Research Paper Available online at: www.ijarcsms.com A Survey:

More information

Router-assisted congestion control. Lecture 8 CS 653, Fall 2010

Router-assisted congestion control. Lecture 8 CS 653, Fall 2010 Router-assisted congestion control Lecture 8 CS 653, Fall 2010 TCP congestion control performs poorly as bandwidth or delay increases Shown analytically in [Low01] and via simulations Avg. TCP Utilization

More information

Applying Router-Assisted Congestion Control to Wireless Networks: Challenges and Solutions 1

Applying Router-Assisted Congestion Control to Wireless Networks: Challenges and Solutions 1 Applying Router-Assisted Congestion Control to Wireless Networks: Challenges and Solutions Jian Pu and Mounir Hamdi Department of Computer Science and Engineering, HKUST, Hong Kong {pujian, hamdi}@cse.ust.hk

More information

High Speed Internet Access Using Satellite-Based DVB Networks

High Speed Internet Access Using Satellite-Based DVB Networks High Speed Internet Access Using Satellite-Based DVB Networks Nihal K. G. Samaraweera and Godred Fairhurst Electronics Research Group, Department of Engineering University of Aberdeen, Aberdeen, AB24 3UE,

More information

Transport layer issues in ad hoc wireless networks Dmitrij Lagutin, dlagutin@cc.hut.fi

Transport layer issues in ad hoc wireless networks Dmitrij Lagutin, dlagutin@cc.hut.fi Transport layer issues in ad hoc wireless networks Dmitrij Lagutin, dlagutin@cc.hut.fi 1. Introduction Ad hoc wireless networks pose a big challenge for transport layer protocol and transport layer protocols

More information

AN IMPROVED SNOOP FOR TCP RENO AND TCP SACK IN WIRED-CUM- WIRELESS NETWORKS

AN IMPROVED SNOOP FOR TCP RENO AND TCP SACK IN WIRED-CUM- WIRELESS NETWORKS AN IMPROVED SNOOP FOR TCP RENO AND TCP SACK IN WIRED-CUM- WIRELESS NETWORKS Srikanth Tiyyagura Department of Computer Science and Engineering JNTUA College of Engg., pulivendula, Andhra Pradesh, India.

More information

Comparative Analysis of Congestion Control Algorithms Using ns-2

Comparative Analysis of Congestion Control Algorithms Using ns-2 www.ijcsi.org 89 Comparative Analysis of Congestion Control Algorithms Using ns-2 Sanjeev Patel 1, P. K. Gupta 2, Arjun Garg 3, Prateek Mehrotra 4 and Manish Chhabra 5 1 Deptt. of Computer Sc. & Engg,

More information

Simulation-Based Comparisons of Solutions for TCP Packet Reordering in Wireless Network

Simulation-Based Comparisons of Solutions for TCP Packet Reordering in Wireless Network Simulation-Based Comparisons of Solutions for TCP Packet Reordering in Wireless Network 作 者 :Daiqin Yang, Ka-Cheong Leung, and Victor O. K. Li 出 處 :Wireless Communications and Networking Conference, 2007.WCNC

More information

Active Queue Management (AQM) based Internet Congestion Control

Active Queue Management (AQM) based Internet Congestion Control Active Queue Management (AQM) based Internet Congestion Control October 1 2002 Seungwan Ryu (sryu@eng.buffalo.edu) PhD Student of IE Department University at Buffalo Contents Internet Congestion Control

More information

TCP, Active Queue Management and QoS

TCP, Active Queue Management and QoS TCP, Active Queue Management and QoS Don Towsley UMass Amherst towsley@cs.umass.edu Collaborators: W. Gong, C. Hollot, V. Misra Outline motivation TCP friendliness/fairness bottleneck invariant principle

More information

First Midterm for ECE374 03/09/12 Solution!!

First Midterm for ECE374 03/09/12 Solution!! 1 First Midterm for ECE374 03/09/12 Solution!! Instructions: Put your name and student number on each sheet of paper! The exam is closed book. You have 90 minutes to complete the exam. Be a smart exam

More information

Outline. TCP connection setup/data transfer. 15-441 Computer Networking. TCP Reliability. Congestion sources and collapse. Congestion control basics

Outline. TCP connection setup/data transfer. 15-441 Computer Networking. TCP Reliability. Congestion sources and collapse. Congestion control basics Outline 15-441 Computer Networking Lecture 8 TCP & Congestion Control TCP connection setup/data transfer TCP Reliability Congestion sources and collapse Congestion control basics Lecture 8: 09-23-2002

More information

Oscillations of the Sending Window in Compound TCP

Oscillations of the Sending Window in Compound TCP Oscillations of the Sending Window in Compound TCP Alberto Blanc 1, Denis Collange 1, and Konstantin Avrachenkov 2 1 Orange Labs, 905 rue Albert Einstein, 06921 Sophia Antipolis, France 2 I.N.R.I.A. 2004

More information

Student, Haryana Engineering College, Haryana, India 2 H.O.D (CSE), Haryana Engineering College, Haryana, India

Student, Haryana Engineering College, Haryana, India 2 H.O.D (CSE), Haryana Engineering College, Haryana, India Volume 5, Issue 6, June 2015 ISSN: 2277 128X International Journal of Advanced Research in Computer Science and Software Engineering Research Paper Available online at: www.ijarcsse.com A New Protocol

More information

Internet Congestion Control for Future High Bandwidth-Delay Product Environments

Internet Congestion Control for Future High Bandwidth-Delay Product Environments Internet Congestion Control for Future High Bandwidth-Delay Product Environments Dina Katabi Mark Handley Charlie Rohrs MIT-LCS ICSI Tellabs dk@mit.edu mjh@icsi.berkeley.edu crhors@mit.edu Abstract Theory

More information

Context. Congestion Control In High Bandwidth-Delay Nets [Katabi02a] What is XCP? Key ideas. Router Feedback: Utilization.

Context. Congestion Control In High Bandwidth-Delay Nets [Katabi02a] What is XCP? Key ideas. Router Feedback: Utilization. Congestion Control In High Bandwidth-Delay Nets [Katabi02a] CSci551: Computer Networks SP2006 Thursday Section John Heidemann 7e_Katabi02a: CSci551 SP2006 John Heidemann 1 Context limitations of TCP over

More information

Seamless Congestion Control over Wired and Wireless IEEE 802.11 Networks

Seamless Congestion Control over Wired and Wireless IEEE 802.11 Networks Seamless Congestion Control over Wired and Wireless IEEE 802.11 Networks Vasilios A. Siris and Despina Triantafyllidou Institute of Computer Science (ICS) Foundation for Research and Technology - Hellas

More information

Using Fuzzy Logic Control to Provide Intelligent Traffic Management Service for High-Speed Networks ABSTRACT:

Using Fuzzy Logic Control to Provide Intelligent Traffic Management Service for High-Speed Networks ABSTRACT: Using Fuzzy Logic Control to Provide Intelligent Traffic Management Service for High-Speed Networks ABSTRACT: In view of the fast-growing Internet traffic, this paper propose a distributed traffic management

More information

Random Early Detection Gateways for Congestion Avoidance

Random Early Detection Gateways for Congestion Avoidance Random Early Detection Gateways for Congestion Avoidance Sally Floyd and Van Jacobson Lawrence Berkeley Laboratory University of California floyd@eelblgov van@eelblgov To appear in the August 1993 IEEE/ACM

More information

First Midterm for ECE374 03/24/11 Solution!!

First Midterm for ECE374 03/24/11 Solution!! 1 First Midterm for ECE374 03/24/11 Solution!! Note: In all written assignments, please show as much of your work as you can. Even if you get a wrong answer, you can get partial credit if you show your

More information

TCP and Wireless Networks Classical Approaches Optimizations TCP for 2.5G/3G Systems. Lehrstuhl für Informatik 4 Kommunikation und verteilte Systeme

TCP and Wireless Networks Classical Approaches Optimizations TCP for 2.5G/3G Systems. Lehrstuhl für Informatik 4 Kommunikation und verteilte Systeme Chapter 2 Technical Basics: Layer 1 Methods for Medium Access: Layer 2 Chapter 3 Wireless Networks: Bluetooth, WLAN, WirelessMAN, WirelessWAN Mobile Networks: GSM, GPRS, UMTS Chapter 4 Mobility on the

More information

Improving XCP to Achieve Max-Min Fair Bandwidth Allocation

Improving XCP to Achieve Max-Min Fair Bandwidth Allocation Improving XCP to Achieve Max-Min Fair Bandwidth Allocation Lei Zan and Xiaowei Yang University of California, Irvine {lzan,xwy}@ics.uci.edu Abstract. TCP is shown to be inefficient and instable in high

More information

Mobile Communications Chapter 9: Mobile Transport Layer

Mobile Communications Chapter 9: Mobile Transport Layer Mobile Communications Chapter 9: Mobile Transport Layer Motivation TCP-mechanisms Classical approaches Indirect TCP Snooping TCP Mobile TCP PEPs in general Additional optimizations Fast retransmit/recovery

More information

THE Transmission Control Protocol (TCP) has proved

THE Transmission Control Protocol (TCP) has proved IEEE TRANSACTIONS ON MOBILE COMPUTING, VOL. 3, NO. 2, APRIL-JUNE 2004 1 Bandwidth Estimation Schemes for TCP over Wireless Networks Antonio Capone, Member, IEEE, Luigi Fratta, Fellow, IEEE, and Fabio Martignon,

More information

Congestion Control Review. 15-441 Computer Networking. Resource Management Approaches. Traffic and Resource Management. What is congestion control?

Congestion Control Review. 15-441 Computer Networking. Resource Management Approaches. Traffic and Resource Management. What is congestion control? Congestion Control Review What is congestion control? 15-441 Computer Networking What is the principle of TCP? Lecture 22 Queue Management and QoS 2 Traffic and Resource Management Resource Management

More information

() XCP-i: explicit Control Protocol for heterogeneous inter-networking November 28th, of high-speed 2006 1 networks / 15

() XCP-i: explicit Control Protocol for heterogeneous inter-networking November 28th, of high-speed 2006 1 networks / 15 XCP-i: explicit Control Protocol for heterogeneous inter-networking of high-speed networks Dino LOPEZ 1 C. PHAM 2 L. LEFEVRE 3 November 28th, 2006 1 INRIA RESO/LIP, France dmlopezp@ens-lyon.fr 2 LIUPPA,

More information

Adaptive Virtual Buffer(AVB)-An Active Queue Management Scheme for Internet Quality of Service

Adaptive Virtual Buffer(AVB)-An Active Queue Management Scheme for Internet Quality of Service Adaptive Virtual Buffer(AVB)-An Active Queue Management Scheme for Internet Quality of Service Xidong Deng, George esidis, Chita Das Department of Computer Science and Engineering The Pennsylvania State

More information

IMPROVING CONGESTION CONTROL FOR END- TO END DELIVERY IN WIRELESS NETWORKS

IMPROVING CONGESTION CONTROL FOR END- TO END DELIVERY IN WIRELESS NETWORKS IMPROVING CONGESTION CONTROL FOR END- TO END DELIVERY IN WIRELESS NETWORKS Satishkumar D. Prajapati 1, Dhaval J. Varia 2 1 PG Student, 2 Assistant Professor, Computer Science and Engineering Government

More information

TCP Startup Performance in Large Bandwidth Delay Networks

TCP Startup Performance in Large Bandwidth Delay Networks TCP Startup Performance in Large Bandwidth Delay Networks Ren Wang, Giovanni Pau, Kenshin Yamada, MY Sanadidi, and Mario Gerla Computer Science Department University of California, Los Angeles Los Angeles,

More information

TTC New Reno - Consistent Control of Packet Traffic

TTC New Reno - Consistent Control of Packet Traffic IMPROVE PERFORMANCE OF TCP NEW RENO OVER MOBILE AD-HOC NETWORK USING ABRA Dhananjay Bisen 1 and Sanjeev Sharma 2 1 M.Tech, School Of Information Technology, RGPV, BHOPAL, INDIA 1 bisen.it2007@gmail.com

More information

15-441: Computer Networks Homework 2 Solution

15-441: Computer Networks Homework 2 Solution 5-44: omputer Networks Homework 2 Solution Assigned: September 25, 2002. Due: October 7, 2002 in class. In this homework you will test your understanding of the TP concepts taught in class including flow

More information

Passive Queue Management

Passive Queue Management , 2013 Performance Evaluation of Computer Networks Objectives Explain the role of active queue management in performance optimization of TCP/IP networks Learn a range of active queue management algorithms

More information

Congestions and Control Mechanisms n Wired and Wireless Networks

Congestions and Control Mechanisms n Wired and Wireless Networks International OPEN ACCESS Journal ISSN: 2249-6645 Of Modern Engineering Research (IJMER) Congestions and Control Mechanisms n Wired and Wireless Networks MD Gulzar 1, B Mahender 2, Mr.B.Buchibabu 3 1 (Asst

More information

TCP for Wireless Networks

TCP for Wireless Networks TCP for Wireless Networks Outline Motivation TCP mechanisms Indirect TCP Snooping TCP Mobile TCP Fast retransmit/recovery Transmission freezing Selective retransmission Transaction oriented TCP Adapted

More information

TCP in Wireless Networks

TCP in Wireless Networks Outline Lecture 10 TCP Performance and QoS in Wireless s TCP Performance in wireless networks TCP performance in asymmetric networks WAP Kurose-Ross: Chapter 3, 6.8 On-line: TCP over Wireless Systems Problems

More information

1. The subnet must prevent additional packets from entering the congested region until those already present can be processed.

1. The subnet must prevent additional packets from entering the congested region until those already present can be processed. Congestion Control When one part of the subnet (e.g. one or more routers in an area) becomes overloaded, congestion results. Because routers are receiving packets faster than they can forward them, one

More information

A Study on TCP Performance over Mobile Ad Hoc Networks

A Study on TCP Performance over Mobile Ad Hoc Networks 215 A Study on TCP Performance over Mobile Ad Hoc Networks Shweta Sharma 1, Anshika Garg 2 1 School of Computing Science and Engineering, Galgotias University, Greater Noida 2 School of Computing Science

More information

ALTHOUGH it is one of the first protocols

ALTHOUGH it is one of the first protocols TCP Performance - CUBIC, Vegas & Reno Ing. Luis Marrone lmarrone@linti.unlp.edu.ar Lic. Andrés Barbieri barbieri@cespi.unlp.edu.ar Mg. Matías Robles mrobles@info.unlp.edu.ar LINTI - Facultad de Informática

More information

Performance Analysis of AQM Schemes in Wired and Wireless Networks based on TCP flow

Performance Analysis of AQM Schemes in Wired and Wireless Networks based on TCP flow International Journal of Soft Computing and Engineering (IJSCE) Performance Analysis of AQM Schemes in Wired and Wireless Networks based on TCP flow Abdullah Al Masud, Hossain Md. Shamim, Amina Akhter

More information

17: Queue Management. Queuing. Mark Handley

17: Queue Management. Queuing. Mark Handley 17: Queue Management Mark Handley Queuing The primary purpose of a queue in an IP router is to smooth out bursty arrivals, so that the network utilization can be high. But queues add delay and cause jitter.

More information

Using median filtering in active queue management for telecommunication networks

Using median filtering in active queue management for telecommunication networks Using median filtering in active queue management for telecommunication networks Sorin ZOICAN *, Ph.D. Cuvinte cheie. Managementul cozilor de aşteptare, filtru median, probabilitate de rejectare, întârziere.

More information

Robust Router Congestion Control Using Acceptance and Departure Rate Measures

Robust Router Congestion Control Using Acceptance and Departure Rate Measures Robust Router Congestion Control Using Acceptance and Departure Rate Measures Ganesh Gopalakrishnan a, Sneha Kasera b, Catherine Loader c, and Xin Wang b a {ganeshg@microsoft.com}, Microsoft Corporation,

More information

Why Congestion Control. Congestion Control and Active Queue Management. Max-Min Fairness. Fairness

Why Congestion Control. Congestion Control and Active Queue Management. Max-Min Fairness. Fairness Congestion Control and Active Queue Management Congestion Control, Efficiency and Fairness Analysis of TCP Congestion Control A simple TCP throughput formula RED and Active Queue Management How RED wors

More information

Analyzing Marking Mod RED Active Queue Management Scheme on TCP Applications

Analyzing Marking Mod RED Active Queue Management Scheme on TCP Applications 212 International Conference on Information and Network Technology (ICINT 212) IPCSIT vol. 7 (212) (212) IACSIT Press, Singapore Analyzing Marking Active Queue Management Scheme on TCP Applications G.A.

More information

Network congestion, its control and avoidance

Network congestion, its control and avoidance MUHAMMAD SALEH SHAH*, ASIM IMDAD WAGAN**, AND MUKHTIAR ALI UNAR*** RECEIVED ON 05.10.2013 ACCEPTED ON 09.01.2014 ABSTRACT Recent years have seen an increasing interest in the design of AQM (Active Queue

More information

Congestion Control for High Bandwidth-Delay Product Networks

Congestion Control for High Bandwidth-Delay Product Networks Congestion Control for High Bandwidth-Delay Product Networks Dina Katabi Mark Handley Charlie Rohrs Ý MIT-LCS ICSI Tellabs dk@mit.edu mjh@icsi.berkeley.edu crhors@mit.edu ABSTRACT Theory and experiments

More information

TCP Adaptation for MPI on Long-and-Fat Networks

TCP Adaptation for MPI on Long-and-Fat Networks TCP Adaptation for MPI on Long-and-Fat Networks Motohiko Matsuda, Tomohiro Kudoh Yuetsu Kodama, Ryousei Takano Grid Technology Research Center Yutaka Ishikawa The University of Tokyo Outline Background

More information

This sequence diagram was generated with EventStudio System Designer (http://www.eventhelix.com/eventstudio).

This sequence diagram was generated with EventStudio System Designer (http://www.eventhelix.com/eventstudio). Client App Network Server App 25-May-13 15:32 (Page 1) This sequence diagram was generated with EventStudio System Designer (http://www.eventhelix.com/eventstudio). TCP is an end to end protocol which

More information

Master s Thesis. A Study on Active Queue Management Mechanisms for. Internet Routers: Design, Performance Analysis, and.

Master s Thesis. A Study on Active Queue Management Mechanisms for. Internet Routers: Design, Performance Analysis, and. Master s Thesis Title A Study on Active Queue Management Mechanisms for Internet Routers: Design, Performance Analysis, and Parameter Tuning Supervisor Prof. Masayuki Murata Author Tomoya Eguchi February

More information

STUDY OF TCP VARIANTS OVER WIRELESS NETWORK

STUDY OF TCP VARIANTS OVER WIRELESS NETWORK STUDY OF VARIANTS OVER WIRELESS NETWORK 1 DEVENDRA SINGH KUSHWAHA, 2 VIKASH K SINGH, 3 SHAIBYA SINGH, 4 SONAL SHARMA 1,2,3,4 Assistant Professor, Dept. of Computer Science, Indira Gandhi National Tribal

More information

Requirements for Simulation and Modeling Tools. Sally Floyd NSF Workshop August 2005

Requirements for Simulation and Modeling Tools. Sally Floyd NSF Workshop August 2005 Requirements for Simulation and Modeling Tools Sally Floyd NSF Workshop August 2005 Outline for talk: Requested topic: the requirements for simulation and modeling tools that allow one to study, design,

More information

TCP/IP Over Lossy Links - TCP SACK without Congestion Control

TCP/IP Over Lossy Links - TCP SACK without Congestion Control Wireless Random Packet Networking, Part II: TCP/IP Over Lossy Links - TCP SACK without Congestion Control Roland Kempter The University of Alberta, June 17 th, 2004 Department of Electrical And Computer

More information

Network Friendliness of Mobility Management Protocols

Network Friendliness of Mobility Management Protocols Network Friendliness of Mobility Management Protocols Md Sazzadur Rahman, Mohammed Atiquzzaman Telecommunications and Networks Research Lab School of Computer Science, University of Oklahoma, Norman, OK

More information

2 TCP-like Design. Answer

2 TCP-like Design. Answer Homework 3 1 DNS Suppose you have a Host C, a local name server L, and authoritative name servers A root, A com, and A google.com, where the naming convention A x means that the name server knows about

More information

Analysis of Internet Transport Service Performance with Active Queue Management in a QoS-enabled Network

Analysis of Internet Transport Service Performance with Active Queue Management in a QoS-enabled Network University of Helsinki - Department of Computer Science Analysis of Internet Transport Service Performance with Active Queue Management in a QoS-enabled Network Oriana Riva oriana.riva@cs.helsinki.fi Contents

More information

TCP Westwood: Congestion Window Control Using Bandwidth Estimation

TCP Westwood: Congestion Window Control Using Bandwidth Estimation TCP Westwood: Congestion Window Control Using Bandwidth Estimation Mario Gerla, M. Y. Sanadidi, Ren Wang, and Andrea Zanella UCLA Computer Science Department Claudio Casetti Politecnico Di Torino Saverio

More information

PART III. OPS-based wide area networks

PART III. OPS-based wide area networks PART III OPS-based wide area networks Chapter 7 Introduction to the OPS-based wide area network 7.1 State-of-the-art In this thesis, we consider the general switch architecture with full connectivity

More information

Analytic Models for the Latency and Steady-State Throughput of TCP Tahoe, Reno and SACK

Analytic Models for the Latency and Steady-State Throughput of TCP Tahoe, Reno and SACK REVISION 1 1 Analytic Models for the Latency and Steady-State Throughput of TCP Tahoe, Reno and SACK B. Sikdar, S. Kalyanaraman and K. S. Vastola Dept. of ECSE, Rensselaer Polytechnic Institute Troy, NY

More information

An Active Network Based Hierarchical Mobile Internet Protocol Version 6 Framework

An Active Network Based Hierarchical Mobile Internet Protocol Version 6 Framework An Active Network Based Hierarchical Mobile Internet Protocol Version 6 Framework Zutao Zhu Zhenjun Li YunYong Duan Department of Business Support Department of Computer Science Department of Business

More information

Performance improvement of TCP over wireless network

Performance improvement of TCP over wireless network Performance improvement of TCP over wireless network Raja singh Computer science Department, SRIT, Jabalpur, M.P.India, rajasinghpatel@gmail.com Brajesh patel Asst. Prof. SRIT,Jabalpur M.P., India, Abstract:

More information

Per-Flow Queuing Allot's Approach to Bandwidth Management

Per-Flow Queuing Allot's Approach to Bandwidth Management White Paper Per-Flow Queuing Allot's Approach to Bandwidth Management Allot Communications, July 2006. All Rights Reserved. Table of Contents Executive Overview... 3 Understanding TCP/IP... 4 What is Bandwidth

More information

Rate-Based Active Queue Management: A Green Algorithm in Congestion Control

Rate-Based Active Queue Management: A Green Algorithm in Congestion Control Rate-Based Active Queue Management: A Green Algorithm in Congestion Control Balveer Singh #1, Diwakar Saraswat #2 #1 HOD Computer Sc. & Engg. #2 Astt. Prof. Computer Sc. & Engg PKITM Mathura (UP) India

More information

TCP Westwood: End-to-End Congestion Control for Wired/Wireless Networks

TCP Westwood: End-to-End Congestion Control for Wired/Wireless Networks Wireless Networks 8, 467 479, 2002 2002 Kluwer Academic Publishers. Manufactured in The Netherlands. TCP Westwood: End-to-End Congestion Control for Wired/Wireless Networks CLAUDIO CASETTI Politecnico

More information

Performance comparison of TCP, HSTCP and XCP in high-speed, highly variable-bandwidth environments

Performance comparison of TCP, HSTCP and XCP in high-speed, highly variable-bandwidth environments Performance comparison of TCP, HSTCP and XCP in high-speed, highly variable-bandwidth environments D. M. Lopez-Pacheco and C. Pham RESO/LIP, University Lyon 1 email:{dmlopezp,congduc.pham}@ens-lyon.fr

More information

Paper C. c Institute of Electrical and Electronics Engineers, Inc. 2000 Reprinted with permission.

Paper C. c Institute of Electrical and Electronics Engineers, Inc. 2000 Reprinted with permission. Paper C Bob Melander, Mats Björkman, and Per Gunningberg. A New End-to-End Probing and Analysis Method for Estimating Bandwidth Bottlenecks. Proceedings of IEEE GLOBECOM 00, San Francisco, CA, USA, November

More information

Final for ECE374 05/06/13 Solution!!

Final for ECE374 05/06/13 Solution!! 1 Final for ECE374 05/06/13 Solution!! Instructions: Put your name and student number on each sheet of paper! The exam is closed book. You have 90 minutes to complete the exam. Be a smart exam taker -

More information

Quality of Service versus Fairness. Inelastic Applications. QoS Analogy: Surface Mail. How to Provide QoS?

Quality of Service versus Fairness. Inelastic Applications. QoS Analogy: Surface Mail. How to Provide QoS? 18-345: Introduction to Telecommunication Networks Lectures 20: Quality of Service Peter Steenkiste Spring 2015 www.cs.cmu.edu/~prs/nets-ece Overview What is QoS? Queuing discipline and scheduling Traffic

More information

CS268 Exam Solutions. 1) End-to-End (20 pts)

CS268 Exam Solutions. 1) End-to-End (20 pts) CS268 Exam Solutions General comments: ) If you would like a re-grade, submit in email a complete explanation of why your solution should be re-graded. Quote parts of your solution if necessary. In person

More information

CS551 End-to-End Internet Packet Dynamics [Paxson99b]

CS551 End-to-End Internet Packet Dynamics [Paxson99b] CS551 End-to-End Internet Packet Dynamics [Paxson99b] Bill Cheng http://merlot.usc.edu/cs551-f12 1 End-to-end Packet Dynamics How do you measure Internet performance? Why do people want to know? Are ISPs

More information

UNBIASED BANDWIDTH ESTIMATION IN COMMUNICATION PROTOCOLS 1

UNBIASED BANDWIDTH ESTIMATION IN COMMUNICATION PROTOCOLS 1 UNBIASED BANDWIDTH ESTIMATION IN COMMUNICATION PROTOCOLS 1 Krister Jacobsson Håkan Hjalmarsson Karl Henrik Johansson {krister.jacobsson hjalmars kallej}@s3.kth.se Department of Signals, Sensors and Systems,

More information

A Spectrum of TCP-Friendly Window-Based Congestion Control Algorithms

A Spectrum of TCP-Friendly Window-Based Congestion Control Algorithms IEEE/ACM TRANSACTIONS ON NETWORKING, VOL. 11, NO. 3, JUNE 2003 341 A Spectrum of TCP-Friendly Window-Based Congestion Control Algorithms Shudong Jin, Liang Guo, Student Member, IEEE, Ibrahim Matta, Member,

More information

Ina Minei Reuven Cohen. The Technion. Haifa 32000, Israel. e-mail: faminei,rcoheng@cs.technion.ac.il. Abstract

Ina Minei Reuven Cohen. The Technion. Haifa 32000, Israel. e-mail: faminei,rcoheng@cs.technion.ac.il. Abstract High Speed Internet Access Through Unidirectional Geostationary Satellite Channels Ina Minei Reuven Cohen Computer Science Department The Technion Haifa 32000, Israel e-mail: faminei,rcoheng@cs.technion.ac.il

More information