<p>A widely studied typical mobile edge computing (MEC) system consists of a cloud server, several edge servers, and some user equipment. Each MEC system is also referred to as a cell. In a multi-cell network, there are several different cells, and the cloud servers in different cells can be connected. In this paper, we consider the problem of offloading computing tasks crossing cells in a multi-cell network to improve the user’s quality of experience (QoE). We first investigate a cross-cell task binary computation offloading and resource allocation model for optimizing QoE and formulate this optimization problem as a mixed-integer nonlinear programming (MINLP). Then, for the offline case, we design an efficient exact algorithm (DGOSS) that can find the optimal computation offloading and resource allocation. For the online case, we devise an online algorithm (DTE-DOL) with a sub-linear bounded regret under dynamic computing task generation, dynamic server quota, and uncertain server-side information assumptions. The online algorithm adopts the multi-user Multi-Armed Bandit technique and distributed auction technique. Finally, we compare the performance of the proposed algorithms on different instances with previously existing algorithms. Experimental results show that for the offline case, the DGOSS algorithm can improve the QoE by approximately 5% with about 37.75% less running time, while for the online case, the DTE-DOL algorithm can significantly improve the QoE by around 18.75% with almost the same running time.</p>