![ The throughput for TCP, which is based on the congestion window behavior, can be
easily found if the cwnd is a constant (flat line) function of RTT. The throughput with
this unrealistic assumption is throughput = cwnd / RTT.
In this assumption, TCP sends a cwnd bytes of data and receives acknowledgement
for them in RTT time.
The behavior of TCP, as shown in Figure, is not a flat line; it is like saw teeth, with
many minimum and maximum values. If each tooth were exactly the same, we could
say that the throughput = [(maximum + minimum) / 2] / RTT.
However, we know that the value of the maximum is twice the value of the
minimum because in each congestion detection the value of cwnd is set to half of its
previous value.
So the throughput can be better calculated as in which
throughput = (0.75) Wmax / RTT
Wmax is the average of window sizes when the congestion occurs.
15
THROUGHPUT](https://image.slidesharecdn.com/tcpcongestion-230514053654-7e0d4dad/85/tcp-congestion-pptx-15-320.jpg)
tcp congestion .pptx
- 1. TCPCONGESTIONCONTROL SUBTITLES INTRODUCTION CONGESTION WINDOW CONGESTION DETECTION CONGESTION POLICIES POLICY TRANSITION TCP THROUGHPUT 1 NETWORKS AND SECURITY
- 2. INTRODUCTION TO TCP CONGESTION 2 When the load offered to any network is more than it can handle,congestion builds up. A router may receive data from more than one sender. No matter how large the buffers of a router may be, it may be overwhelmed with data, which results in dropping some segments sent by a specific TCP sender. In other words, there is no congestion at the other end, but there may be congestion in the middle. TCP needs to worry about congestion in the middle because many segments lost may seriously affect the error control. More segment loss means resending the same segments again, resulting in worsening the congestion, and finally the collapse of the communication.
- 3. 3 TCP is an end-to-end protocol that uses the service of IP. The congestion in the router is in the IP territory and should be taken care of by IP. However, as we discussed before, IP is a simple protocol with no congestion control. TCP, itself, needs to be responsible for this problem. TCP cannot be very conservative, either, sending a small number of segments in each time interval, because this means not utilizing the available bandwidth of the network. TCP needs to define policies that accelerate the data transmission when there is no congestion and decelerate the transmission when congestion is detected. TCP uses different policies to handle the congestion in the network. They are, Slow Start Additive Increase Fast Recovery
- 4. CONGESTION WINDOW To control the number of segments to transmit, TCP uses another variable called a congestion window (cwnd), whose size is controlled by the congestion situation in the network . The cwnd variable and the rwnd variable together define the size of the send window in TCP. The first is related to the congestion in the middle (network); the second is related to the congestion at the end. The actual size of the window is the minimum of these two, Actual window size = minimum (rwnd, cwnd) 4
- 5. CONGESTION DETECTION 5 The TCP sender uses the occurrence of two events as signs of congestion in the network: time-out and receiving three duplicate ACKs. The first is the time-out. If a TCP sender does not receive an ACK for a segment or a group of segments before the time-out occurs, it assumes that the corresponding segment or segments are lost and the loss is due to congestion. Another event is the receiving of three duplicate ACKs (four ACKs with the same acknowledgment number). Recall that when a TCP receiver sends a duplicate ACK, it is the sign that a segment has been delayed, but sending three duplicate ACKs is the sign of a missing segment, which can be due to congestion in the network. A very interesting point in TCP congestion is that the TCP sender uses only one feedback from the other end to detect congestion. Time-out, is the sign of a strong congestion; the receiving of three duplicate ACKs is the sign of a weak congestion in the network.
- 6. CONGESTIONPOLICIES TCP’s general policy for handling congestion is based on three algorithms: SLOW START (Exponential Increase) ADDITIVE INCREASE (Congestion Avoidance) FAST RECOVERY
- 7. SLOW START cwnd = 1 2⁰ cwnd= cwnd + 1 = 1 + 1 = 2 2¹ cwnd= cwnd + 2 = 2 + 2 = 4 2² cwnd= cwnd + 4 = 4 + 4 = 8 2³ In the slow-start algorithm, the size of the congestion window increases exponentially until it reaches a threshold.
- 8. ADDITIVE INCREASE size of the congestion window in this algorithm is also a function of the number of ACKs that have arrived and can be determined as follows: If an ACK arrives, cwnd = cwnd+(1/cwnd). The size of the window increases only 1/cwnd portion of MSS (in bytes). In other words, all segments in the previous window should be acknowledged to increase the window 1 MSS bytes.
- 9. FAST RECOVERY Like congestion avoidance, this algorithm is also an additive increase, but it increases the size of the congestion window when a duplicate ACK arrives (after the three duplicate ACKs that trigger the use of this algorithm). The fast-recovery algorithm is optional in TCP. The old version of TCP did not use it, but the new versions try to use it. It starts when three duplicate ACKs arrive, which is interpreted as light congestion in the network.
- 10. POLICY TRANSITION We discussed three congestion policies in TCP. Now the question is when each of these policies policies is used and when TCP moves from one policy to another. To answer these questions, we need to refer to three versions of TCP: Taho TCP Reno TCP New Reno TCP
- 11. TAHO TCP: The early TCP, known asTaho TCP used only two different algorithms in their congestion policy: slow start and congestion avoidance.
- 12. RENO TCP: A newer version of TCP, called Reno TCP, added a new state to the congestion-control called the fast-recovery state.
- 13. NEWRENO TCP: 13 A later version of TCP, called NewReno TCP, made an extra optimization on the Reno TCP. In this version, TCP checks to see if more than one segment is lost in the current window when three duplicate ACKs arrive. When TCP receives three duplicate ACKs, it retransmits the lost segment until a new ACK (not duplicate) arrives. If the new ACK defines the end of the window when the congestion was detected, TCP is certain that only one segment was lost. However, if the ACK number defines a position between the retransmitted segment and the end of the window, it is possible that the segment defined by the ACK is also lost. NewReno TCP retransmits this segment to avoid receiving more and more duplicate ACKs for it.
- 14. ADDITIVEINCREASE,MULTIPLICATIVEDECREASE: 14 The congestion window size, after it passes the initial slow-start state, follows a saw tooth pattern called additive increase, multiplicative decrease (AIMD).
- 15. The throughput for TCP, which is based on the congestion window behavior, can be easily found if the cwnd is a constant (flat line) function of RTT. The throughput with this unrealistic assumption is throughput = cwnd / RTT. In this assumption, TCP sends a cwnd bytes of data and receives acknowledgement for them in RTT time. The behavior of TCP, as shown in Figure, is not a flat line; it is like saw teeth, with many minimum and maximum values. If each tooth were exactly the same, we could say that the throughput = [(maximum + minimum) / 2] / RTT. However, we know that the value of the maximum is twice the value of the minimum because in each congestion detection the value of cwnd is set to half of its previous value. So the throughput can be better calculated as in which throughput = (0.75) Wmax / RTT Wmax is the average of window sizes when the congestion occurs. 15 THROUGHPUT
- 16. Example: If MSS = 10 KB (kilobytes) and RTT = 100 ms in Figure 24.35, we can calculate the throughput as shown below. Soln: Wmax = (10 + 12 + 10 + 8 + 8) / 5 = 9.6 MSS Throughput = (0.75 Wmax / RTT) = 0.75 × 960 kbps / 100 ms = 7.2 Mbps 16