802.11無線網路之分散式負載平衡演算法

在過去十年間，IEEE 802.11無線區域網路在校園、家庭和企業中變得越來越普及。就因為它的普及化及方便，每個人皆可利用來存取無線網路。然而當越來越多人加入無線網路中，一些IEEE 802.11連結準則的缺點就顯露了出來。換句話說，當最大接收訊號強度指數(RSSI)準則被廣泛的運用時，使用者的負擔常常不均勻的分佈在眾多的無線接取點(Access Point)中。在IEEE 802.11的原始設計，當一個工作站(STA)有資料必須傳送，則該工作站必先競爭頻道的使用權。而且當有持續的壅塞情形發生，競爭視窗(contention window)則會呈現指數的增加。因此當有大量使用者時，如何針對IEEE802.11無線網路來設計、佈建，來減輕壅塞的情形和分散負擔在眾多的無線接取點中是非常重要。近年來，許多研究專注於把工作站的個數當作成負載指數(load index)。一些研究則把全部流量(throughput)當成該無線接取點的負載指數。然而因為無線網路在實體層(PHY layer)變動的特性，這些負載指數皆不能確實的反應一個無線接取點的負載。在這篇碩士論文中，我們藉由在競爭中的延遲時間(defer time)和平均傳送虛耗(transmission overhead)提出了一個新的負載指數。其中傳送虛耗可用一個機率模型來計算。以該負載指數為基礎，我們也提供了一個具啟發性、分散性的負載平衡演算法。在我們的提案中，工作站可動態的允許跟一個無線接取點作連結，而且該分散式演算法也可被所有的無線接取點所執行，在一個廣大和擁塞的無線網路環境中以達到更好的負載平衡。為了驗證我們提出的負載指數和演算法效能，我們寫了個C語言模擬程式，並以網路電話(VoIP)和網頁(Web)當作我們的分析應用程式。模擬結果顯示該提議的負載指數和負載平衡演算法可以增進網路電話的品質，且全部的網頁存取流量可全面的增加。

關鍵字

負載平衡演算法；無線網路

並列摘要

In the past decade, IEEE 802.11 wireless local network has been more and more popular in most campus, home, and enterprise. Many people can access wireless network because its popularization and convenience. However, some disadvantages of association criterion in IEEE 802.11 are revealed when there are more and more users join in the wireless network. In other words, user load is often unevenly distributed among Access Points (APs) when the maximum signal strength criterion is widely used for selecting an AP. In original design of IEEE 802.11, every station (STA) must contend for the channel usage if it has data to transmit. And the contention window would be increased by an exponential law if collisions continue to occur. Therefore, how to alleviate the congestion situation and distribute load among all APs is critical for those IEEE 802.11 wireless network designed and deployed for large number of users. In recent years, many studies focus on the number of STAs associated with an AP as the load index of the AP. Some research works also take the total throughput of an AP as the load index. However, the load indexes can not reflect the accurate load of an AP because of the fluctuating characteristic in the physical layer of the wireless network. In the thesis, we propose a new load index which takes into account of the defer time during the contention and the average transmission overhead evaluated via a probabilistic model. Based on the load index definition, we then propose heuristic distributed load-balancing algorithm. In the proposal, STAs are allowed to associate with an AP dynamically and the distributed load-balancing algorithm can be executed by all APs to reach better load balancing in a large and congested WLAN environment. In order to verify the performance of the proposed load index and load-balancing algorithm, we execute the simulation program written in C language, and employ VoIP and Web traffic as the analyzing network applications. The simulation results show that the proposed load index and load-balancing algorithm can improve the VoIP quality, and the total web access throughput is improved overall.

並列關鍵字

load balance algorithm ； wireless

參考文獻

[22] S.-T. Sheu and C.-C. Wu, “Dynamic Load Balance Algorithm (DLBA) for IEEE 802.11 Wireless LAN,” in Tamkang Journal of Science and Engineering, vol. 2, No. 1 pp. 45-52, 1999.

[2] Y. Bejerano, S.-J. Han, L. Li, “Fairness and load balancing in wireless LANs using association control,” Proc. of 10th annual international conference on Mobile computing and networking, Philadelphia, PA, USA, pp. 315-329, Sep 2004.

[3] H. Velayos, V. Aleo, G. Karlsson, “Load balancing in overlapping wireless LAN Cells,” Communications, 2004 IEEE International Conference, pp. 3833-3387 Vol. 7, June 2004

[4] A. Balachandran, P. Bahl, G. M. Voelker, “Hot-spot Congestion Relief in public-area Wireless Networks,” in Proc. of 4th IEEE Workshop on Mobile Computing Systems and Applications, June 2002

[5] D. Liu,F. Huebner, “Application profiling of IP traffic,” Local Computer Networks, 2002, 27th Annual IEEE Conference, Nov. 2002.

國際替代計量

802.11無線網路之分散式負載平衡演算法

全文下載

主題瀏覽