Stochastic Analysis of Delayed Mobile Offloading in Heterogeneous Networks

Mobile cloud offloading that migrates heavy computation from mobile devices to powerful cloud servers through communication networks can alleviate the hardware limitations of mobile devices thus providing higher performance and saving energy. Different applications usually give different relative importance to response time and energy consumption. If a delay-tolerant job is deferred up to a given deadline, or until a fast and energy-efficient network becomes available, the transmission time will be extended, which can save energy because a more energy-efficient communication channel and a less energy-restricted computation platform may become available. However, if the reduced service time fails to cover the extra waiting time, this policy may not be competitive. In this paper, we investigate two types of delayed offloading policies, the partial offloading model where jobs can leave from the slow phase of the offloading process and be executed locally on the mobile device, and the full offloading model, where jobs can abandon the WiFi Queue and be offloaded via the Cellular Queue. In both models, we minimize the Energy-Response time Weighted Product (ERWP) metric. Not surprisingly, we find that jobs abandon the queue often when the availability of the WiFi network is low. In general, for delay-sensitive applications the partial offloading model is preferred under a suitable reneging rate, while for delay-tolerant applications the full offloading model shows very good results and outperforms the other offloading model when selecting a large deadline. From the perspective of energy consumption, the full offloading model will always be best, even if the deadline must be extremely long. Only if job response time is of high importance an optimal deadline to abort offloading in the partial offloading model or the WiFi transmission in the full offloading model can be found. For reduction of the energy consumption it will always be better to wait longer rather than compute locally or use the cellular network.