Due to the limitation of the mobile device battery lifespan, the concept of Mobile Cloud Computing(MCC) offloading is popularly introduced to handle the conflicts between the resource-hungry mobile tasks and the limited internal energy capacity. To further reduce the communication delay and energy cost between the offloader and offloadee, the Cloudlet model is proposed that performs as the agent of remote Cloud Datacenter. In this case, many offloading tasks used to be executed by the distant Cloud can be processed locally on the Cloudlet. However, unlike the remote Cloud that is assumed to provide ”unlimited” computing utility, the Cloudlet is still bounded regarding computing power, storage, network bandwidth and coverage. In this paper, a Cloudlet-based offloading model is proposed to enable energy and execution efficient offloading. A task-centric resource allocation model is presented to handle the resource limitation issue of the local Cloudlet. In the experiments, the proposed model is compared to the traditional device-based solutions, and the offloading execution results present an overall improvement of the offloading energy reservation as well as execution throughput.