Statistical quality of service provisioning over edge computing mobile wireless networks

The edge computing techniques have been developed to support the exponentially increasing service demands in the fifth generation (5G) networks by bringing the data contents and their corresponding computations/communications near to mobile users. As a promising and efficient solution to implement edge computing techniques, mobile data offloading complements the traditional cellular transmission through either WiFi networks or device-to-device (D2D) communication/relay to mitigate the overloaded wireless traffics for cellular networks. However, the mobile data offloading in edge computing wireless networks imposes many new quality of service (QoS) guarantee problems which still remain opening research areas. To overcome these challenges, in this paper we propose the statistical delay-bounded QoS provisioning schemes using the effective capacity theory for two types of mobile data offloading: WiFi offloading and D2D offloading, where we formulate the WiFi offloading and D2D offloading as the single-hop and two-hop wireless-link statistical QoS provisioning problems, respectively. We also derive the optimal probability of using each type of offloading which maximizes the average effective capacity for all mobile users under the following two scenarios: (1) all mobile users have the same probability of requesting the popular multimedia data contents and have the same probability to become a relay in D2D offloading; and (2) each mobile user has the distinct probability of requesting the popular multimedia data contents and the distinct probability to become a relay in D2D offloading. Finally, we evaluate and validate our proposed statistical delay-bounded QoS provisioning schemes over multimedia data offloading architecture through numerical analyses.