High Performance Computing for Cyber Physical Social Systems by Using Evolutionary Multi-Objective Optimization Algorithm

TLDR

Cyber‑physical social systems combine cyberspace, physical space, and social space, and many of their problems are multi‑objective optimization tasks that are difficult to solve with traditional methods, necessitating high‑performance computing. The study applies evolutionary multi‑objective optimization algorithms to address multi‑objective optimization problems in cyber‑physical social systems. A floorplanning case study demonstrates the approach, employing a B*-tree representation and a multistep simulated annealing algorithm in tandem. Experimental results show the method can find feasible floorplans with a 74.44 % success rate (268 out of 360 trials).

Abstract

Cyber-physical social systems (CPSS) is an emerging complicated topic which is a combination of cyberspace, physical space, and social space. Many problems in CPSS can be mathematically modelled as optimization problems, and some of them are multi-objective optimization (MOO) problems (MOPs). In general, the MOPs are difficult to solve by traditional mathematical programming methods. High performance computing with much faster speed is required to address these issues. In this paper, a kind of high performance computing approaches, evolutionary multi-objective optimization (EMO) algorithms, is used to deal with these MOPs. A floorplanning case study is presented to demonstrate the feasibility of our proposed approach. B*-tree and a multistep simulated annealing (MSA) algorithm are cooperatively used to solve this case. As per experimental results for this case, the proposed method is well capable of searching for feasible floorplan solutions, and it can reach 74.44 percent (268/360) success rates for floorplanning problems.