隨著物聯網(IoT)、大數據(Big data)與雲端(Cloud)等科技的興起,無線感測器通訊網路提供居家智慧家庭一個可行的平台,其特色在於將智慧家庭中各式各樣的感測器利用無線網絡的方式,集中於一個集線器(Hub)後再一起上傳雲端伺服器。相較於直接讓所有的感測器具備連上網的功能,此方法能有效地降低上網的功耗(power consumption)。本篇論文實現了一個從底層到上層的一個完整無線傳輸系統設計,包括了物理層(Physical layer)與應用層(Application layer)的設計。與傳統的設計方式不同的是,由於我們同時在設計時考慮了不同層(Layer)間的需求,可藉此達到優化整體系統的目的,而非優化單獨區塊後再串接。同時,整體性的優化也會衡量介面的問題(interface problem),並藉由調整上層或下層的設計來解決。除此之外,為了能達到系統性地最佳化整體無線感測器通訊網路,我們提出了一個簡化的數學模型,模擬整體網路(network)的使用情境,並藉由此模型評估整系統就功耗與傳輸效率(transmission efficiency)層面的表現。之後再利用成本函數(Cost function)與功耗每字節參數 (Energy Per Byte Index)找出最佳的設計參數。藉由上述這些方法,我們能利用具備系統性且高效率的方式來設計無線感測器通訊網路,並提供低功耗的系統架構以配合物聯網低功耗的需求趨勢。
As the Internet of Thing (IoT) trend, cloud technology, and big data analysis technology grow, the wireless sensor network is introduced for the in-house IoT scenario. In this thesis, we design both the communication protocol and the transmission hardware together to optimize the overall system performance. Due to the cross-layer design, the interface problem becomes an important issue. Besides, a systematic analysis of a simple network architecture is conducted to find the optimal operational point of the protocol design, given the hardware parameters. In other words, a model considers the clock drift, BER vs SNR, the synchronization error, the sync. frequency, the sync. margin, fail packet statistics and so on, is proposed as a guideline for wireless communication network design. Moreover, a cost function in algorithm field and the Energy Per Byte Index (EPBI) are introduced to evaluate the performance in the perspective of power consumption V.S. data rate. Most important of all, most of the parameters in the analysis are based on the measurement result and hence are much more realistic for the integrated system design in practical. In short, a design rule for the wireless sensor network design is proposed and an optimal protocol parameter set in the perspective of power consumption is found for a fixed specification of the adopted hardware, which meets the low power consumption trend of the IoT application.