具備低能量消耗之公平獎勵的整合策略

Abstract

叢集演算法常被使用來產生長時效的網路拓樸，因為這些算法的隨機性和動態調整，收發封包的工作量會被整個網路的節點所均攤。然而，真正使能耗降低的因素是資料的整合，從而使傳輸與接收的資料量降低。並且，資料整合策略可以確保大部分的資料在等待整合的過程中，不至於超過其資料的有效時間。大部分的資料整合策略在設計時，沒有考慮到網路的拓樸或路由結構，但是這兩者跟資料整合策略的效能與參數是高度相關的，如離終點有幾個中繼站和接收封包的速度。 本論文提出一個新的基於叢集算法特性的資料整合策略，以達到更好的能源效率以及更低的資料超時率。透過預測資料超時的情況，我們的方法分析並計算即將過期與獲得的資料，最後決定傳輸的時間點。實驗模擬的結果顯示，我們花在傳輸的電力比第二好的算法低上大約10%到40%，並且大部分只有0.5%到5%的資料過期率。

Clustering algorithms are the most common methods to create long lifetime network topologies. Due to the dynamic nature and randomness of clustering algorithms, the workload of transmission and reception can be amortized by different nodes. However, the main idea behind saving energy is that data aggregation compression can reduce the data to transmit, and the data aggregation policy is to ensure that the most data can be aggregated without being expired. Most of the data aggregation policy discusses their mathematical model without concerning topology and routing protocol, but yet the topology and routing is closely related to data aggregation policy performance and its parameter, such as number of hop to the data sink and rate of incoming packets. This paper proposes a new data aggregation policy utilizing the features of clustering algorithms to better improve energy efficiency and expiration rate. By predicting the expiration of data, our method calculates and compares between the number of expiring and incoming data to decide the moment of transmission. The simulation shows that our transmission energy is 10% to 40% lower than the second best solution and most of the packet drop rate is about 0.5% to 5%.