MAS computing method for dynamic bandwidth distribution

Abstract

The invention discloses a MAS computing method for dynamic bandwidth distribution. The method comprises the following steps: clearly defining user level UL of each user, arrival time AT of a data packet and a data packet length PL; clearly defining the weight alpha of the user level, the weight beta of the data packet arrival time AT, and the weight gamma of the data packet length; clearly defining a comprehensive effect estimation parameter Ui(pi) of the user pi; and obtaining a minimum value of a target function Si of the ONU Agent in a polling period. By adopting a modelling method of the MAS, the Agent modeling is performed on the service in the OLT and the ONU, the related factors influencing the bandwidth distribution is abstracted, the local bandwidth redistribution behavior of thebargaining based on the IPACT polling game principle is designed, the bandwidth distribution is benefited according to the economic view, the queuing problem in the data packet transmission process iswell solved, the transmission time slot is eliminated, and the USR elimination effect is enhanced.

Description

technical field [0001] The present invention relates to dynamic bandwidth allocation, in particular to the issue of bandwidth allocation under the joint influence of multiple factors such as user level, data packet, and time delay, and in particular to a multi-agent system (multi-agent system, MAS) calculation for dynamic bandwidth allocation method. Background technique [0002] In the traditional bandwidth allocation process, the Optical Line Termination (OLT) demodulates the data obtained from the network and sends it to each Optical Network Unit (ONU), and the ONU demodulates the received information and sends it to each user. Under this mechanism, data transmission is linear and sequential. When multiple influencing factors act on different applications, it cannot effectively solve the queue sorting problem. There is mutual influence between services, plus user level, delay, data packet When factors such as size are intertwined, it will result in inefficient allocatio...

AI Technical Summary

Problems solved by technology

Under this mechanism, data transmission is linear and sequential. When multiple influencing factors act on different applications, it cannot effectively solve the queue sorting problem. There is mutual influence between services, plus user level, delay, data packet When factors such as size are intertwined, it will cause inefficient allocation

The algorithm used by OLT and ONU to participate in bandwidth allocation is interleaved polling with adaptive cycle time (IPACT). The calculation of this algorithm is in bytes, but the queue is scheduled in packets It is carried out as a unit, which causes the phenomenon that the bandwidth is vacant because it is less than the length of a data packet, resulting in a large unused small overhead (unused slot remainder, USR), resulting in a waste of bandwidth resources

Images