QoS Wi-Fi網路系統之分群傳送實作

Abstract

Wi-Fi採用IEEE 802.11的標準而生，在無線區域網路當中佔著相當重要的地位，由於在無線網路上其媒介為共用，因此必須要有能控制媒介存取的機制，在Wi-Fi中採用的媒介控制機制為分散式協調功能(Distribution Coordination Function, DCF)，其講求保障所有裝置在長期來看能有相同的傳輸機會，使得裝置在相同速率下可以有平等的性能表現。但是隨著技術的進步，實體層上的傳輸速率越來越快也越多，此時整體系統的輸出將受限於DCF的機制，由於強調裝置之間平等的傳輸機會，突然出現少量的低傳輸速裝置時，DCF機制將造成整體傳輸效能的驟減，這被稱為性能異常 (performance anomaly)。為了解決這個問題，先前研究者提出了獨立不同速率的裝置，將其依速率區分為不同群組個別配置傳輸時間這樣的做法，裝置只被允許在其所屬群組可傳輸的區間進行傳輸，如此一來不同速率的裝置將不互相影響並擁有相同的媒介使用時間，解決媒介分配的有效性與公平性問題。然而其研究的論文中只有理論，尚未將其實踐，因此在我們的研究中，借用了其概念並略做修改，並在實體環境上將實踐，取名為基於時間平等之Wi-Fi系統 (Time-based fairness Wi-Fi, TF-Wi-Fi)，實踐了控制裝置之傳輸行為，並建立了群組管理系統，將裝置劃分為不同群組，並分配不同群組間的傳輸時間，以解決多重速率環境下的媒介分配問題，最後透過不同的實驗中觀察到，當高、低傳輸速率相差極大時，TF-Wi-Fi能大幅提升高速裝置與整體效能。

Wi-Fi based on IEE 802.11 standard plays an important role in wireless network. In wireless network, the medium is shared and therefore there must be a mechanism to control the access of medium. In Wi-Fi, Distribution Coordination Function (DCF) used as the mechanism of Medium Access Control (MAC) grantees the fairness of transmission opportunity among stations. However, with evolutions of wireless communication technology, the transmission rate becomes much higher and variety. The fairness of transmission opportunity in DCF is not efficient to distribute the resource of medium access anymore. In multi-rate environment, the performance of stations with high transmission rate is dramatically decrease once a few stations turns into low transmission rate. This phenomenon is named performance anomaly. To resolve this problem, previous researcher proposed an idea to isolate the stations with high transmission rate and low transmission rate. All stations are divided into serval groups depending on its transmission rate and can only transmit when the group it belongs to is permitted to access the medium. Therefore, the interference between stations with high transmission rate and low transmission rate is eliminated. In our research, we modify that idea and build a realistic Wi-Fi system called Time-based Fairness Wi-Fi (TF-Wi-Fi). TF-Wi-Fi realizes the transmission control enforcement on stations and constructs a clustering management system to divide the stations into groups and distribute the medium resource. Finally, we conduct serval experiments and observe a dramatic improvement of throughput when variance of transmission rates between stations are prodigious.