Computation Offloading in Mobile Cloud Computing

Mobile Cloud Computing is an offspring of Cloud Computing and Mobile Computing in which Computation Offloading plays a very important role. Computation Offloading is the process of transferring heavy compute- intensive applications from the mobile phone to the public cloud as the mobile phones are not able to cater the needs of such heavy applications. Different algorithms were formulated to offload the heavy applications efficiently like Deep Reinforcement Learning Computation Offloading Algorithm and Artificial Bee Colony Optimization Algorithm. The Reinforcement algorithm is used so to determine as to which host the application can be migrated without overloading it and avoiding deadlocks. Deep Reinforcement Learning Computation Offloading Algorithm is a machine learning algorithm which uses the reward based technique to evaluate as to which virtual machine the application should be offloaded so that it is definitely executed. The Artificial Bee Colony Algorithm is efficiently applied for offloading massive applications from mobile phones to the cloud and is cross validated using Convolution Neural Network until last application. Convolution Neural Network is similar to the Neural Networks and have a specific kind of layers. Each layer of a Convolution Neural Network transforms a 3 Dimensional input volume to a 3 Dimensional output volume of neuron activations. Every offloaded task is cross verified using Convolution Neural Network. The Convolution Neural Network uses confusion matrix to give accurate results. Different parameters like Efficiency, Computational Intensity, Resource Utilization and Computation Overhead are formulated to give the accuracy of Computation Offloading using Deep Reinforcement Learning Computation Offloading Algorithm and Artificial Bee Colony Algorithm and also Convolution Neural Network. A SWOT (Strengths, Weakness, Opportunities and Threats) analysis of different Computation offloading techniques, is made. Computation Offloading uses the technique to migrate heavy computational resource applications from smart-phone device to the cloud. Emphasis is laid on the similarities and differences of computation offloading algorithms and models. Computation Offloading parameters include the user’s context, performance context and resource context awareness. This work focus on single site, dynamic and virtual machine computation offloading which is application and user specification based.