CSMA/CA 無線網路上結合模擬退火與基因演算法調整載波偵測門檻值之協定設計

Abstract

隨著IoT的發展以及無線設備的與日俱增，使得網路環境變得日益壅塞，如何更有效的分配資源，讓通道使用率發揮到最大值是眾所關注的議題。通道的使用率與傳輸的可靠度為一體兩面，增加同時可傳輸的裝置能提升使用率，但同時也增加碰撞的風險，降低傳輸的可靠度。因此，如何在提升通道使用率與維持傳輸可靠度之間取得平衡即為本篇論文的主旨。

在CSMA/CA的無線網路架構下，調整物理載波偵測門檻(Physical Carrier Sensing Threshold)為一有效且向後兼容(backward compatible)的機制，提升偵測門檻能有效地提升空間重複使用率，進而提升網路效能。但過多的同時傳輸則會大幅提升碰撞風險，因而使網路效能降低。本篇論文所提供的演算法，致力於在該環境下選出物理載波偵聽門檻的最佳值。融合基因演算法(Genetic Algorithm)及模擬退火法(Simulated Annealing)的優勢，在合理的計算時間內，選出最佳值使網路效能最佳化，在模擬中亦驗證了我們的假設。

With the arising of Internet of Things (IoT) applications and the fast growing of wireless devices, the Internet has become more and more crowded. How to deploy the resources effectively and make the best use of wireless channel is an essential yet non-trivial problem. Generally, increasing the spatial reuse will decrease the channel reliability simultaneously. In CSMA/CA-based wireless networks, adjusting physical carrier sensing threshold is an effective way to influence the spatial reuse factor. Increasing carrier sensing threshold typically encourages a better spatial reuse, which also means a better networks capacity. However, too many simultaneous transmissions will incur a higher packet collision probability. In this paper, we propose to combine the Simulated Annealing and Genetic Algorithm to obtain a global network-wise carrier sensing threshold for throughput improvement. Our simulation results validate the efficacy of the proposed tuning approach.