Research And Design Of Distributed Task Management System For Mobile Application Detection

With the development of mobile Internet, people have become more and more dependent on these mobile applications. However, due to the lack of supervision from store, the attract of illegal interest and the difference of developers’ level, more and more malicious applications appeared, which caused to many security events. In order to solve this problem, we need to detect application in quantities before they are admitted to store. We prefer distributed system for parallel detection than single-machine detection because of the variety of detection technology and the large number of applications. After analyzing the current distributed task management framework, it is found that none of them are suitable for Mobile application detection tasks. Therefore, this paper designs a distributed task management system for mobile application detection.This paper studies the current mobile application detection technology and analyzes the advantages and disadvantages of them firstly. Then this paper studies the current distributed task management framework and analyzes the using limit and design point of them. Based on all the studies, we summary the characteristics of mobile application detection task and propose a detection method which contains a variety of detection methods in the form of work-flow in order to finish detection more quickly. Then the key technologies of distributed task management system are introduced.In this paper, DAG is used to describe rapid detection, and the task scheduling strategy based on dynamic feedback is used to schedule task ,and Zookeeper is used to solve the single-point problem of centralized management nodes, and the TFS is used to store massive small file. In the end, this paper designs and implements a distributed task management system, proposes functional requirements and performance requirements, and introduces the system architecture and system flow. This papers describes the design and implementation of four function modules in detail, including platform client, centralized management layer, detection engine layer and data storage layer.In this paper, the system topology and system construction method are given. The feasibility of the distributed task management system is verified by experiment. Through the performance testing, the effect of the scheduling strategy is analyzed. The average detection time of detection task is tested and the recovery time of single point crash is also tested. It is proved that the designed system is efficient, highly available and extensible, and meets all the requirements of security detection of mobile applications.