適用於多參數無線感測器網路之具高適應性無失真資料壓縮演算法設計

網路壽命的延長是無線感測器網路技術現今重要的研究議題。無線傳輸為無線感測器節點最為耗能的運作程序，因此透過資料壓縮可以降低無線傳輸資料量，進而延長網路壽命。隨著無線感測器網路的應用發展日趨多元化，多感測參數的應用情境逐漸受到重視，因此資料壓縮演算法亦必須將多參數的適用性納入設計考量。有鑑於此，為滿足網路應用的需求，本研究提出一套適用於多參數無線感測器網路之具高適應性無失真資料壓縮演算法(High-Adaptive Lossless Data Compression Algorithm for Multi-parameter Wireless Sensor Networks, HALCAM) HALCAM對於每一項感測參數進行獨立的數值預測，可以獲得較佳的預測誤差，並且僅須傳輸因此相對較小的預測誤差至接收端，故HALCAM具備良好的適應性；除此之外，各項感測參數的預測誤差會進行更進一步地資料壓縮，HALCAM透過一混合基底記數系統，將預測誤差以簡短的方式表示，其基底依照近期數據的標準差進行動態調整，進而獲得較好的壓縮效率。本研究為了測試HALCAM的壓縮效率，使用不同標準差及機率分佈的模擬樣本進行分析，相較於前人的研究，當模擬樣本的標準差提升時，HALCAM依然可維持良好的壓縮效率；接著，本研究亦引用來自公開資料庫的真實樣本進行驗證，其呈現與模擬樣本相同的結果。本研究為了探究HALCAM的延遲時間與節能效率的，引用熱門感測器節點的產品規格進行分析實驗。實驗結果證實，當無線感測器網路的感測參數項數為2~5項時，HALCAM的延遲時間為480~1552 μs，而節能效率為10~40%；其中兩項指標皆隨著項數提升，說明HALCAM可以為感測參數日趨多元化的網路應用提供更優異的壓縮效率與節能效率。

關鍵字

多參數無線感測器網路；無失真資料壓縮演算法；高適應性

並列摘要

Network lifetime prolongation is a critical issue for wireless sensor networks (WSNs). As radio communication consumes most power, the extension of network lifetime can be generally achieved by data compression which reduces the data transmission and reception of a radio transceiver. Moreover, with the rapid development of WSN applications, the applicability of data compression in multi-parameter networks is becoming important. In order to satisfy these requirements, a high-adaptive lossless data compression algorithm for multi-parameter wireless sensor networks (HALCAM) has been proposed. HALCAM is greatly adaptive by combining a prediction method for each monitoring parameter individually, and only the prediction errors need to be transmitted. Furthermore, the prediction errors are represented with a mixed radix system, and the radixes are determined by the standard deviations of moving windows dynamically. In this study, the compression ratio of HALCAM is evaluated by compressing simulated datasets with varied standard deviations and probability distributions. Comparing with an existing algorithm, the result shows that HALCAM can maintain greater compression ratio when the standard deviations of the datasets increase. The proposed algorithm is also validated by other tests with data acquired from real WSNs. In addition, some commercial sensor nodes are employed to examine the performance of HALCAM in time delaying and energy saving. The simulate results show that HALCAM can save 10~40% energy and only costs 480~1552 μs when compressing 2~5 monitoring parameters.