Stability Analysies Of TCP/RED Congestion Control Systems

In traditional Drop-Tail routers, packets are dropped only when buffers are full. This approach has many drawbacks. It creates volatile and unstable queue length that causes long, variable and unpredictable packet queuing delay and creates periods during which the system over-reacts to congestion signals that lead to link underutilization.One way to overcome the drawbacks of Drop-Tail is to use the so-called active queue management (AQM). Unlike Drop-Tail, AQM drops packets before a queue becomes full so that a source can respond to congestion before buffers overflow. One AQM, the so-called random early detection (RED) has been recommended by the IETF. The main design objectives of RED are to eliminate unnecessary packet loss, increase link utilization and decrease queuing delays. RED achieves these goals by introducing an enhanced control mechanism involving randomized packet dropping and queue length averaging.By applying the time-delay control theory to a TCP/RED dynamic model, this paper establishes some explicit conditions under which the TCP/RED system is stable in terms of the averaged queue length. Then, the stability condition and the relation between the control parameter maxp of RED and the network parameters are discussed. Finally, the results are illustrated by using ns2 simulations, which demonstrates that it is able to choose appropriately maxp based on the stability conditions derived in this paper, to achieve satisfactory network performance. It is found, by comparison, that this improved performance is better than that of three other typical AQM schemes - Proportional-Integral (PI) controller, the Random Exponential Marking (REM), and Adaptive Virtual Queue (AVQ) schemes.