Abstract

This thesis entails the development of concurrent algorithms and specific architectures for the Protocol Data Unit (PDU) encoding/decoding functionality of communication protocol implementations. The main motivation in the development of these algorithms and architectures is to increase the end-to-end performance of multilayered communication architectures where high computational costs associated with PDU encoding/decoding directly affect the performance. In order to construct the formal basis of this development, the notions of type and value of PDUs are initially defined. Based on the type-value duality and different forms of values, PDU encoding and decoding are divided into phases, which are then identified as mappings between different value forms. Founded on this formalism, concurrent algorithms for PDU encoding/decoding are developed. A distributed implementation model for these algorithms is then conceptualized, where the Communicating Sequential Processes (CSP) notation is used as the basis of formalism to specify the model. Using the distributed implementation model, a taxonomy for different architectural models for PDU encoders/decoders is presented. A customized performance evaluation methodology for measuring the effectiveness of the architectures is proposed. The overall applicability of all the above concepts is demonstrated through the development, implementation, and performance evaluation of an Abstract Syntax Notation-One (ASN.1) encoder/decoder used in Open System Interconnection (OSI) implementations.