Heterogenous Server Placement for Delay Sensitive Applications in Green Mobile Edge Computing

The importance of environmental pollution has motivated the researchers to investigate the electrical power consumption of any emerging technology. Edge Computing (EC) technology, with a significant number of power-hungry computing servers for accepting the offloaded intensive computing tasks from mobile devices, is no exception. It provides a method to reduce cloud communication latency by creating a small-scale cloud computing paradigm in close proximity to users. But, to reduce power consumption, it is necessary to minimize the number of servers used in the edge environment. For this purpose, server placement and load balancing are two important factors that must be taken into account. Previous researches have revealed the impact of server placement on the power consumption of network edge equipment. However, in addition to power consumption, it is also important to consider the delay of the offloaded task. One issue that has received less attention in server placement is queueing delay. Compared to other network delays (such as transmission and propagation delays), queueing delay is of particular importance due to its random nature. In this paper, using the M/G/m queueing model for the edge servers, an optimization problem is formulated in order to reduce the power consumption of the edge environment. The proposed formulation considers queueing delay in addition to server placement and load assignment. It is also worth mentioning that the presented model is assumed to benefit from an overlay network infrastructure equipped by Software Defined Network (SDN) technology that helps to balance the load between servers and decrease the overall delay. Simulation results demonstrate the effectiveness of the proposed solution.