Control of Traffic Congestion with Weighted Random Early Detection and Neural Network Implementation

摘要

Applying Quality of Service mechanisms to modern communications is essential for the efficiency and for the traffic reliability. The various Quality of Service methods are based on queues management depending on individual traffic parameters. Choosing Quality of Service parameters on the edge network devices defines the management queue and packet discard/queued parameters on the intermediate devices. The proposed research explores the possibility of automatically adapting to the already selected class based Quality of Service policy, of new users added to the backbone of the network. A neural network is trained to automaticlly adapt new end users to the quality of service policy, already set by other end-users and accepted by the intermediate routers. The obtained results show that the automated adaptation of the Quality of Service parameters to the already set ones, is possible for the intermediate routers, and the positive consequences of applying such a method are mentioned.