A Stable Stateless Fair Bandwidth Allocation Algorithm Using Stochastic Control

Because of their simplicity and low expense, the study of stateless fair bandwidth allocation algorithm became a hot spot among various AQM algorithms. However, among current stateless fair bandwidth allocation algorithms, the UDP flow´s bandwidth is still far from the theoretic fair bandwidth, the protection for TCP flows and the throughput stability are insufficient. A stateless fair bandwidth allocation algorithm using stochastic control -P-CHOKe algorithm was proposed. This scheme selects multiple packets from the queue, the more packets have the same flow ID with the incoming packet means the larger sending speed these flows have. Packet is dropped with probability according to the packet proportion. Simulation results show that compared to M-CHOKe and rubbish bin algorithm, this algorithm can achieve fairer bandwidth allocation, and have better ability to identify UDP flow. It can effectively reduce TCP packet dropping rate. In the mean time, this algorithm has high throughput stability and it can achieve smooth throughput of the flows