一個具高度可調適性與雙門檻值的RPL目標函數設計分析

Abstract

在無線感測網路中存在著各式各樣的應用，不同的應用之間透過網路所表現之特性可能存在著極大的差異。在一個多跳網路中如何選擇下一跳節點會對最終形成的網路拓樸的特性產生決定性的影響。在無線感測網路中多數應用採用電池供電，因此有效率的使用電源在無線感測網路中是一個極為重要的課題。本篇論文中，我們提出一個能夠有效的在能源消耗與效能之間進行調整的高度可調適的RPL目標函數。此目標函數可以根據特定的應用對各個所需的路由度量進行分級，並針對每一個特定的路由度量的各層級適當地設置一個數值來影響節點在下一跳節點選擇上的考量。這樣的設計不僅能夠在不影響其他路由度量的情況下針對新的應用快速的新增它所需的特殊路由度量。並且在我們的設計下能夠更精確地去排除鏈路狀況未達一定需求的節點，以達到在符合應用所需的效能需求情況下盡可能地去延長網路地生命週期。這樣的設計使得我們的目標函數能夠更有效的滿足各種應用的需求。最後，我們經由Cooja模擬器和現存的方法(OF0和MRHOF)進行比較以評估我們的方法之效能，由模擬結果可以看出我們的方法能夠大幅地延長整個網路的生命週期，並且仍然能夠保持良好的封包接收率與封包延遲時間。

There are a lot of diverse applications in wireless sensor networks (WSNs). Requirements of these applications might be quite different. How to select the best neighbor to be the preferred parent will greatly affect the performance of final network topology. Most of the devices are battery-powered in wireless sensor networks, thus energy efficiency is a key issue in WSN. In this thesis, we proposed an objective function, Highly Adjustable Dual Thresholds Object Function (HADTOF), which accounts energy consumption and link quality to be the main factor and the second factor, respectively for routing decision. In our proposed scheme, we classify each routing metric to multiple levels and in each level we can use different rank calculation formula to calculate the rank increment value. In this rank calculation mechanism, when the quality of a specified routing metric becomes poor, we will add a penalty value to make the candidate parent become worse than the others. And the rank increment of each routing metric will accumulate to reflect the comprehensive score of the candidate parent. With this mechanism, we can quickly adjust our objective function to make it appropriate to another different application. (No matter adding a new routing metric or adjusting the original routing metric considerations to avoid some situations which must be avoided in the routing path.) Finally, we evaluate the performance of our proposed scheme by simulation and compare it with OF0 and MRHOF. The result shows that our approach can extend network lifetime substantially while still keep high delivery ratio and low end-to-end delay.