Research On Compressive Sensing Based Data Collection In Wireless Sensor Network

One of the critical issue of Wireless sensor network(WSN) is how to reduce the energy consumption during data gathering. Data compression is one of an efficient way to prolong the network lifetime. Most traditional data compression methods are not suitable for wireless sensor network due to it needs large data exchanges or the complexity computation. Recently, Compressive Sensing(CS) opened up a new research avenue due to it has accurately recover, and it simplifies the process of data encoding. The research is shown as below:First, this paper does the different studies on the spatial-temporal data gathering models. To illustrate the restoration precision is influenced by different reconstruction algorithm and measurement matrix, this paper analysis the incoherence of sensing matrix and does variety experiments on data reconstruction.Second, this paper focuses on the combination of cluster routing protocol and sparse projection. To reduce the energy consumption while gathering observing values, this paper combine CS with LEACH and put forward a new approach using remaining power feedback to solve the problem that sensors have different energy consumption due to they have different communication distance with cluster head. In the proposed approach, sensors can dynamically adjust their sampling according to the average remaining power.Finally, this paper proposed a self-adapting sampling method. The data reconstruction error cannot be guaranteed due to the data's sparsity is dynamically changing over time. To solve this problem, this paper presents a novel approach for sensors to dynamically adjust their sampling rate by analyzing the tendency of data. In order to validate the proposed approach, this article implement this scheme on Crossbow MicaZ platform.