Pervasive wireless sensor network for real-time environmental information rendering.

摘要

We envision wireless sensor networks that support the rendering of real-time information about an environment. In particular we focus on the effectiveness of wireless senor networks in the sensing of temperature distribution in a fire scene. Since tiny sensors are energy-constrained and inaccurate, the challenge is to reduce the overall transmission energy cost and improve the data accuracy through distributed data compression and noise reduction. Tackling this problem can be achieved by taking advantage of the redundancy in the densely populated and randomly spaced sensor data. We propose an energy-efficient Alpha Tree Routing Algorithm and a data collection framework that consists of two distributed data compression schemes and one local error correction algorithm.;Telescopic Data Compression is proposed for its fast and energy-efficient scanning of a phenomenon over a large region. It does the scan through multi-resolution sampling with in-cluster low-pass filtering and a progressive zoom-in process. Complementary to Telescopic Data Compression is Grid-Based Haar Compression. This compression scheme allows stricter control on reconstruction accuracy through progressive trimming of details in a bottom-up hierarchical approach.;Telescopic Data Compression is first used to identify certain areas of interests that demand higher accuracy. The Grid-Based Haar Compression is then applied in those regions to capture the finer details. To further boost up the reconstruction accuracy, we use the Distributed Feedback Algorithm to enhance the quality of individual sensor data through iterations of local feedbacks.;Simulation results based on the temperature distribution of a building fire generated by the NIST Fire Dynamics Simulator show that with a sensor density of 1.2 /m2 and a SNR of 6 dB. Using Telescopic Data Compression alone, it produced an overall average improvement of 4.4 dB. This can be obtained while collecting only 4.8% of the raw data traffic at the processing center. If Grid-Based Haar Compression is applied instead, the SNR gain is raised to 5.4 dB by sending 26% of data. Distributed Feedback Algorithm can bring a prominent SNR gain of 7.9 dB on individual sensor data. If Telescopic Data Compression is then applied on the feedback-enhanced data, the SNR gain becomes 5.8 dB. With Grid-Based Haar Compression, the SNR gain is up to 7.1 dB while keeping 15% of data. Crucial information about a fire can be effectively extracted from the reconstructed field.;With our generic data collection framework, a sensor network application can flexibly adjust its data collection strategy based on the relative information content of different regions: from fast and crude scanning for large-scale signal overview, to the extraction of details at high accuracy in specific areas of interests. By smartly balancing the tradeoff between signal reconstruction accuracy, compression ratio and speed of response, energy is more effectively spent, thus maximizing the overall lifetime of the network.