This study explores distributed computing techniques in the context of dynamic network algorithms for intelligent transportation systems (ITS) applications. ITS technologies such as Advanced Traveler Information Systems (ATIS) and Advanced Traffic Management Systems (ATMS) provide traffic-related information on-line for a new generation of models and methodologies that aim at the realtime enhancement of network performance and efficiency. However, existing sequential processing solution techniques and/or single-processor computational environments preclude translation of these methodologic and algorithmic advances to on-line operability due to the severe computational burden of processing networkwide time-dependent traffic-related data (such as volume, speed, occupancy, and classification). To address this problem, we investigate two remote procedure call (RPC)-based distributed computing techniques on a network of workstations. Computational results indicate that the distributed implementation performs better than sequential computation, although tradeoffs between communications overheads and computational savings become critical with the number of processors. This suggests a threshold number of processors for optimal computational performance depending on the specific problem and its solution logic. The results also illustrate the effects of congestion levels, loading profiles, and output data size on computational performance.
