A Computational Task Offloading Method Based on Differential Evolution in Mobile Cloud Environment

Abstract

The invention discloses a computing task unloading method based on differential evolution in a mobile cloud environment, the steps of which are: transforming the inferred computing process into a task graph, constructing a task unloading model; measuring the similarity of population individuals, and obtaining the initialization population with the largest difference; Weighted fusion population evolution algebra and individual fitness adjust the scaling factor, and select the mutation strategy according to the scaling factor; mix the target individual and the variable individual dimension components to generate a cross individual, compare its fitness with the target individual, and retain the individual with better fitness to enter The next generation; according to the population fitness variance to measure the aggregation degree of population individuals, randomly select some individuals for secondary mutation; judge whether the number of iterations is satisfied, if so, output the code of the optimal individual in the population, otherwise continue to iterate; The encoding of the individual is decoded into a task offloading solution, and the solution is output. The algorithm of the invention has strong optimization ability, and can effectively shorten the task response time under the condition of satisfying the cost constraint.

Description

technical field [0001] The invention belongs to the field of mobile cloud computing, in particular to a computing task offloading method based on differential evolution in a mobile cloud environment. Background technique [0002] With the development of mobile devices and embedded devices, deep learning computing also needs to be applied on these devices. CNN is an important branch of deep learning and has been widely used in speech recognition, document analysis, language detection and image recognition. As a computing-intensive network, CNN generally adopts cloud computing solutions, but this limits the applications that require real-time response, and may cause privacy leakage and increased energy consumption. In recent years, the number of CNN layers has continued to deepen, and mobile devices with small storage space and limited computing power cannot meet the storage and computing requirements of CNN models. Therefore, it is also facing great difficulties to directly ...

AI Technical Summary

Problems solved by technology

[0003] Purpose of the invention: In order to overcome the problems in the prior art, aiming at the problem that the offloading strategy with the single goal of minimizing time delay is difficult to meet the user’s QoS requirements, the present invention provides a computing task offloading method based on differential evolution in a mobile cloud environment, It can improve the algorithm optimization ability, effectively shorten the task response time and make unloading decisions adaptively according to the set cloud platform cost constraints and different network speeds

Images