Search

Much progress has recently been made to enhance mesoscopic traffic simulation with the focus on advanced modeling features and traffic management support capabilities. Subsequently, the computational performance needs to be improved in order to make mesoscopic traffic simulation stay effective to real-time applications. This paper presents a framework for mesoscopic traffic simulation which offloads both the demand and supply components to GPU. The simulation algorithm divides the simulation flow into different steps and designs multiple kernels to handle the steps. A high level of data parallelism is achieved by assigning the GPU threads to the appropriate components of the traffic network. Several optimization options using an innovative data structure and improved warp execution are also deployed to harness the GPU performance while preserving simulation correctness. The performance of the framework is evaluated in a real network showing the speedup of up to nearly 5 times in the demand simulation and more than 4 times in the supply simulation compared to the sequential simulation. [ABSTRACT FROM AUTHOR]

Mesoscopic traffic simulation is a promising technology for handling large-scale dynamic traffic assignment problems. Meanwhile, GPU has been a research hotspot in the last decade because of its massive parallel performance compared with CPU. Although several works deal with GPU-based parallel performance, few have addressed the issue of real-world network applications. Thus, the performance of GPU on realistic road networks is still not clear. In this paper, an ideal grid network and a realistic Singapore expressway network are implemented and compared. The former gets a speedup of 10.00 while the latter only get a speedup of 2.37. The two main reasons for this lower efficiency of a real network are the more complex data structure caused by network topology and fewer numbers of threads due to the scale of the expressway network. Future works need more effort to develop a system with better data structure and better designed concurrent threads. [ABSTRACT FROM AUTHOR]