Dynamic Transmission Rate Allocation in Packet Networks Using Recorrent Neural Networks Trained with Real Time Algorithm
Flávio Henrique Teles Vieira, Rodrigo Pinto Lemos, Luan Ling Lee · IEEE Latin America Transactions · 2003
In this paper recurrent neural networks are considered to realize traffic prediction in computer network. The transmission rate that must be allocated in order to prevent byte losses and to get an efficient network use is estimated in real time. For such, recurrent neural networks were trained with real time learning algorithms: RTRL (Real Time Recurrent Learning) and extended Kalman filter. The algorithms are applied in the dynamic transmission rate allocation in a network link, verifying its efficiencies in the traffic prediction and control.