Path selection enabling user mobility and efficient distribution of data for computation at the edge of mobile network

Convergence of mobile networks and cloud computing enables to offload heavy computation from a user equipment (UE) to the cloud. The offloading can reduce energy consumption of the UEs. Nevertheless, delivery of data to a centralized cloud leads to high latency and to overloading backhaul network. To overcome these constrains, computing capabilities can be brought closer to the user and integrated into small cell base stations deployed in mobile networks. This concept of cloud-enabled small cells is known as small cell cloud (SCC). In the SCC, the UEs benefit from proximity to the computing stations resulting in both lower latency and alleviating load of backhaul. In this paper, we propose a path selection algorithm finding the most suitable way for data delivery between the mobile UE and the cells performing computation for this particular UE. The path selection algorithm estimates transmission delay and energy consumed by the transmission of offloaded data and selects the most suitable base station for radio communication accordingly. The path selection problem is formulated as Markov Decision Process (MDP). The algorithm is suitable for parallel computation in dynamic scenarios with mobile users and handles mobility for users exploiting computing services in the SCC. Comparing to conventional approach for delivery of data to computing cells, the proposed algorithm reduces the delay up to 54.3% and UE's energy consumption is decreased by up to 7.5%. Moreover, users' satisfaction with data transmission delay is increased by up to 28% and load of small cell's backhaul is lowered by up to 29%.