Research And Implementation On Task Scheduling Model For Multi-Agent Heterogeneous System Based On Pi-Calculus

The distributed system has become a popular research object in the computer field with its excellent computing power and excellent scalability.It can collect data from decentralized hosts and utilize these hosts to complete the calculation of big data,which involves to the task scheduling problem in distributed systems in calculation process.The current distributed systems are gradually becoming heterogeneous and the task scheduling problem in heterogeneous distributed systems is NP-complete,so how to quickly and efficiently complete task scheduling in heterogeneous distributed systems has become the main research direction in distributed field.Pi-calculus is mainly used in the study of concurrency models in distributed systems.It has a strong ability to simulate the communication topology,build channels between distributed nodes to complete the process migration and offer a correctness verification of the model.It is an indispensable mathematical basic formalism for describing and analyzing concurrent computing in distributed systems.Agents have the characteristics of autonomy,initiative,and sociality.The communication delay and communication cost between hosts in a heterogeneous distributed system can be reduced through the cooperation and interaction between agents,and data can be migrated with agents,thereby reducing the network load.Based on the research of heterogeneous distributed systems,task scheduling,Pi-calculus and Agent technology,this paper proposes a multi-agent task scheduling model for heterogeneous system based on Picalculus,namely PMTS model.The main work of this paper is as follows:First,propose a PMTS model by integrating Pi-calculus and Agent technology into the task scheduling of heterogeneous distributed systems.This model uses Pi calculus to improve the coordination and communication efficiency between different hosts,enhance the concurrency between different hosts and within a single host,then uses Agent technology to enhance the computing performance of heterogeneous distributed systems and improve task allocation.It can face more uncertainty in task scheduling of heterogeneous distributed systems,and is more flexible and efficient.Second,build the PMTS model,verify its correctness and implement it.The process in Pi-calculus is used to represent the active entities in the PMTS model,and channels in Pi-calculus are used to interact between various active entities.The MWB verification tool is used to complete the deadlock detection of mobile process on PMTS model and the weak bisimulation theory of Pi-calculus is used to complete the equivalence proof of PMTS model.Then,the PMTS model in this paper is implemented by using Nomadic Pict,a concurrent programming language with Pi calculus as the core.Third,use DAGs with different topologies to simulate tasks,do horizontal and vertical comparison experiments on PMTS models.The JADE-based task scheduling model and PMTS model are compared horizontally,then Pi-calculus-based task scheduling model without Agents and PMTS model are compared vertically.Multiple sets of experimental data are given and results are compared and analyzed.It is verified that the PMTS model proposed in this paper can complete task scheduling faster and improve the efficiency of task scheduling in heterogeneous distributed systems by comparing the results of experiments,which is of great significance to research direction of task scheduling in heterogeneous distributed systems.