Internet of Things (IoT) is the major research filed in the recent trends. IoT has the ability to create communication with any object. IoT produces big amount of raw data at the time of data gathering. Therefore, there is a need of efficient mechanism to address the issue of IoT. This paper presents the localized MapReduce Framework for IoT. This frame work processes the data at the local nodes without transferring the data to the cloud or high end servers. The mobile agents have the ability to migrate from one node to another node for data processing, Mapper and Reducer are used for agent duplication and result aggregation. The performance of the proposed framework is evaluated using the cost function and the results proved the efficiency the proposed framework.