Research And Design Of Program Components Division Strategy For Mobile Computation Offloading

With the rapid development and coverage of mobile Internet,mobile intelligent terminal become the hottest communication and entertainment equipment.The applications have a greater demand for memory and storage space,and even at the same time use terminal sensors?camera?microphones and other large energy-consuming components.Obviously,the size of the mobile terminal and the battery has been growing,but only improving hardware technology has been far from meeting the needs of the user's energy consumption.Therefore,reducing the energy consumption of the terminal operation becomes a thorny problem.Now,the computation offloading has become the key means to reduce the energy consumption of the terminal.Computation offloading is to migrate part of calculation of terminal complex applications to a higher-speed server to perform,and will enforce the results back to the migration point,thus reducing the terminal running time and energy consumption.The most important research problem in computation offloading is how to divide the program correctly,which can reduce the terminal energy consumption.The paper focuses on the problem of program division,proposing an adaptive wireless network environment program component division strategy based on the existing research.There are three main aspects of the program division problem to be solved.First,abstract correct program running relationship model;second,according to the running environment,correctly divide components;third,according optimization goal,making the best offloading decision.1)For the problem of abstracting the operation relationship model of program components,the paper based on the knowledge of graph theory,components are respectively abstracted as chain structure and graph structure according to the structure and component operating relationship.Among them,the vertex weight represents the component calculation,while the side right represents the component data interaction.For applications that cannot be directly abstracted as a chain structure,a component relationship transformation algorithm is proposed.The algorithm can ensure the correct execution of the flow while transforming the relationship model,which can reduce the complexity of program division.2)For the problem of program division,in this paper,we propose two adaptive network algorithms based on the two component models,which are the chain program division algorithm and the complex program segmentation algorithm.The proposed chain component division algorithm based on the actual network environment by looking for the best starting position and ending position to get the minimum energy consumption value and carry out the offloading decision.The complex program component division algorithm reduces the component number and complexity of the application by a coarsening step in Multi-level algorithm,we design a new coarsening utility function as the coarsening criterion,then according to actual network environment to dynamically adjust offloading unit.Two algorithms are verified by simulation experiments with low time complexity,and can adjust division result according to the variation of network bandwidth.3)For the program of optimal target and migration decision,the paper uses the necessary condition of computation offloading saving energy consumption combined with the Shannon formula to get the relationship between network bandwidth,the amount of calculation and the amount of communication,as a division algorithm model.One more,we propose the optimal energy consumption target formula in the time delay constraints.