Optimal Joint Power Allocation and Task Splitting in Wireless Distributed Computing

The mobile application market is growing very fast. More and more applications require intensive computations. However, computation power of mobile devices is limited and does not catch up with the growth of computation demand of applications. Computation offloading is a promising approach for reducing the computation load and reducing execution time and energy consumption. In this work, we investigate a wireless distributed computing scenario where a mobile device exploits the computation resources of the nearby devices to reduce execution time. The main goal is to study the gain of computation offloading among heterogeneous devices. We formulate the problem of optimal offloading as a joint resource allocation problem consisting of power allocation and task splitting. The resulting problem is non-linear and non-convex. By exploiting the unique characteristics of the problem, we can transform it into a convex problem. The optimal solution of the original problem can then be determined by performing a bisection search over one parameter. Using numerical simulation, we can show that by offloading computation load to nearby devices, the execution time is reduced substantially. Moreover, the gain of offloading in comparison with computing locally increases with the ratio of the computation load to the size of input data of the task as well as with increasing number of nearby devices.