Abstract: Neuron PID algorithm can control queue length successfully, but its neuron gain is sensitive for controlled object state. So it is difficult to guarantee the control performance because the gain depends on experience and trial method. Congestion window size, loss probability and queue length differential equations were deduced based on TCP window additive-increase and multiplicative-decrease (AIMD)principle and queuing mechanism. TCP/AQM control model was obtained in Ad-hoc network through the equations lincarization. Then it introduced recursion and modification gain to neuron PID based on the model. Finally, a neuron adaptive proportional summation differential (PSD) AQM was proposed. PSdalgorithm can modify neuron gain dynamically. NS simulations demonstrate that PSD ctueue management performance is better than neuron PID under conditions of wireless packet loss,sudden flow and different link capacity.

Key words: Ad-hoc network, Congestion control, AQM，PSD