Research on Routing Strategies of Wireless Sensor Networks with Applications on Environmental Monitoring

由於無線感測器具微小化之特殊需求，且須長時間持續以無線方式進行感測與傳輸，如何於有限電量供應下，使此類感測器組成之網路壽命可盡量延長，便是無線感測器網路系統研究所面臨之重要課題。其中，由於無線傳輸部份乃是現今無線感測器耗能之一大來源，若能有效根據現有的無線電技術，發展出合適的傳輸機制，以節約網路中各感測器之耗電量，便可有效達成此目標。 此類傳輸機制之一：負載平衡，乃藉由將網路中因資料傳輸需求所造成之能量消耗分散至網路中各感測器，盡可能使網路各部份之能量消耗(負載)平均分配，以延長無線感測器網路壽命之技術。本研究藉探討現有負載平衡技術之特點，而根據一個雙步作法提出以多樹架構延伸 Ad Hoc On-Demand Vector Routing 進行分層之演算法。藉由對多樹架構中各路徑賦予一權重，部署後之網路便可動態選擇多樹之一進行資料回傳，以達到負載平衡目的。本文據此演算法分別提出集中式與分散式策略，並針對網路規模於策略優劣之影響進行分析，以瞭解如何不同網路規模選擇適合策略。 藉由電子計算機模擬，可由多個參數瞭解此策略之成效。採用本文提出策略之網路，比起現今常見之決定式策略，部署完成後之負載平衡狀態較佳。此策略所需之參數，亦可藉由此類模擬進一步瞭解，進而針對個別網路部署環境所需進行設定。It is important to extend the lifetime of wireless sensor networks (WSNs) as long as possible with limited energy supplies, as sensors in these networks are built with minimized sizes, and are required to sense and wirelessly transmit data for long periods. Because wireless transmission is one of the more dominant sources of energy consumption for sensors used in WSNs, the above objective can be reached if a suitable transmission strategy is developed on the basis of radio transmission technologies of today. One of the strategies—load balancing—is to reduce network hot spots by spreading inevitable energy consumption (loads) across the wireless sensors in the network. This thesis introduces a two-step solution to this problem based on the Ad Hoc On-Demand Vector Routing tier allocation with multiple paths to disperse loads from hot spots. After the allocation, paths for nodes to transmit data back to the sink are decided through a probabilistic approach based on weights of trees. Based on the algorithm structure, two both centralized and distributed strategies are developed, and the suitableness of them are analyzed against network sizes. The performance of the algorithm is shown via various indices after computer simulations. Compared to common deterministic approaches, the strategy proposed is able to generate a more load-balanced topology with network deployment, and the network built reaches an appropriate state in reasonable time. Simulations are also done to map variables dominant to the selection of strategies against common parameters controllable during network setup