Parallel processing scheme of the received data based on OSI protocols

Abstract

This paper studies a parallel processing scheme for the OSI protocols in session layer and below to achieve high-speed processing for the received data. First, technical issues are extended to exploit parallelism based on the layered structure of the OSI reference model. A basic parallel processing model including the distribution function for the received protocol data units (PDUs) and the synchronization mechanism is presented to make each microprocessor assigned to a layer execute the protocol processing for the received data in parallel. Next, the configuration of the distribution function is discussed with a particular emphasis on reassembling of segmented PDUs. Then a distribution processing scheme is proposed and algorithm is selected from the consideration on the separation processing corresponding to the concatenation of PDUs. Here, the issues are examined relating to execution of the proposed algorithm, such as the prevention of the disorder of the data unit sequence induced by the reassembling process. Finally, the characteristics of the parallel processing scheme are presented based on the proposed algorithm by comparing the communication performance with that of the ordinary sequential processing scheme by computer simulations.