Design And Implementation On Code Injection And Memory Tracking In Dalvik Virtual Machine

With the rapid development of wireless network, a large number of mobileterminal devices which are represented by smartphone start to access network. Manyaccess methods, such as Internet,3G and wireless broadband network show the trend ofconvergence. Ubiquitous network era which is full of networks and applications is beinggradually realized. As a mobile terminal device, which is easier to carry andcommunication, Android phones are loved by broad masses. However, due to itshardware limitations, Android phones’ computing power and storage capacity arelimited.Cloud computing is a virtual computing resource pool. What’s more, it is able todynamically layout application environment. As cloud computing real-time monitorsapplication resource usage, it will allocate resources to the application which is lack ofresources.Cloud computing parallel processes data and services on a distributedenvironment and provides large-scale data storage. Overall, cloud computing providessafer storage and efficiency operation.If we combine cloud computing with mobile computing, we will be able to get thedesired results by using high-speed processing capability of remote cloud even thoughthe phone performs low. With the existing transmission bandwidth and mobile terminallimited computing power, we need to balance the load banlancing of the two sidesrather than put all calculations to the cloud. So in order to achieve an optimal efficiencyupgrade, how to allocate the terminal application?The author transfer heavy computation mobile applications to the cloud, so that themobile applications’ efficiency will be greatly improved. The biggest lightspot isseamlessly transferred the heavy computation mobile application to the cloud.This paper takes function running process and memory management in Dalvikvitual machine as the main research field, deeply analysis and research the keytechnique in it. The author proposes a method to automatically transfer function to thecloud. The work on this paper can be described as follows:First, propose one plan to improve function’s execution process. This plan adds several interfaces and methods in Java layer, JNI layer and Native layer so that thefunction can be transferred to run in the cloud instand of Dalvik vm.Second, propose an analysis mode about function memory. The mode analyzesfunction’s total memory data, such as register, function stack frame, heap, stack and soon. Put forward different improved method to get different data. By using the analysismode memory data to create a new operating environment in the cloud, function can runnormally.We find through experiment results of Anrdoid simulator and Androiddevelopment board that the efficiency of the implementation of the application will beincreased by at least20times by transfering heavy computation mobile applications tothe cloud. So people can use Android mobile phones anywhere and anytime to access,get and use many services, such as prospect extraction, face recognition and picturefilter provided by cloud computing quickly and easily.