Pradyut Nath, Sumagna Dey, Subhrapratim Nath, A. Shankar, J. Sing, Subir Kumar Sarkar
{"title":"VLSI Routing Optimization Using Hybrid PSO Based on Reinforcement Learning","authors":"Pradyut Nath, Sumagna Dey, Subhrapratim Nath, A. Shankar, J. Sing, Subir Kumar Sarkar","doi":"10.1109/VLSIDCS53788.2022.9811434","DOIUrl":null,"url":null,"abstract":"Rapid advances in Very Large-Scale Integration (VLSI) technology demand wire length minimization of the circuits in VLSI physical layer design to ensure routing optimization. With the growing dimension of the circuits and increased complexity, only transformation of VLSI routing problem into Non-Polynomial (NP) complete Rectilinear Minimal Spanning Tree (RMST) problem and solving it with traditional approaches results in non-optimal solutions. This brings the need for metaheuristic algorithms. Using metaheuristic algorithms, finding the optimal placement of Steiner points by approximation became easier to optimize the routing path, but sometime with major deviation. In this proposed work, A hybrid Particle swarm optimization (PSO) is used which optimizes and estimates using a value Iteration matrix, obtained using Reinforcement Learning (RL). This RL guided PSO generates much better solutions safely and with more consistency when compared with existing metaheuristic-based routing algorithms. The collected findings demonstrate that the proposed methodology has a lot of potential in VLSI routing optimization.","PeriodicalId":307414,"journal":{"name":"2022 IEEE VLSI Device Circuit and System (VLSI DCS)","volume":"36 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2022-02-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2022 IEEE VLSI Device Circuit and System (VLSI DCS)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/VLSIDCS53788.2022.9811434","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
Abstract
Rapid advances in Very Large-Scale Integration (VLSI) technology demand wire length minimization of the circuits in VLSI physical layer design to ensure routing optimization. With the growing dimension of the circuits and increased complexity, only transformation of VLSI routing problem into Non-Polynomial (NP) complete Rectilinear Minimal Spanning Tree (RMST) problem and solving it with traditional approaches results in non-optimal solutions. This brings the need for metaheuristic algorithms. Using metaheuristic algorithms, finding the optimal placement of Steiner points by approximation became easier to optimize the routing path, but sometime with major deviation. In this proposed work, A hybrid Particle swarm optimization (PSO) is used which optimizes and estimates using a value Iteration matrix, obtained using Reinforcement Learning (RL). This RL guided PSO generates much better solutions safely and with more consistency when compared with existing metaheuristic-based routing algorithms. The collected findings demonstrate that the proposed methodology has a lot of potential in VLSI routing optimization.