一個在無線區域網路上增進VoIP傳輸效能的適應性輪詢機制

Abstract

隨著網路電話，以及無線網路系統的逐漸普及，將兩者結合為無線網路電話也將是未來的必然新趨勢。隨著這個趨勢的發展，無線網路標準IEEE 802.11e便孕育而生了，它保證了在無線網路上的服務傳輸品質，這對於語言的傳遞這種敏感度較高的應用尤其重要。它提供了一種新的介面傳輸方式命名為HCF，它用來解決IEEE 802.11中舊有的DCF以及PCF。為了要減少傳輸時間的延遲以及在各無線裝置間的輪巡的時間，我們提出了一個新的輪巡方式，它是在HC端以HCCA的模式下運行，其中，HC中會將所有與之相連的無線設備分為兩類並將之分別記錄起來，其中一類為運作中的設備，另一類為非運作中的設備，這兩筆記錄除了會隨著目前的狀況做動態的更新，還會依據HC和各無線裝置的傳輸方向和等候列中的情況，去做最正確的記錄和排列。我們提出了一個簡單的方式去判定無線裝置是否是在運作的狀態，這種方法我們稱為VBR和CBR。運作中設備的TXOP是和非運作中設備是分開做排程的。而後，HC會在這兩大類設備的記錄中去做輪巡。從模擬的結果可以清楚的看到，輪巡所花費的時間，已經減少很多了，因此我們可以得到多放傳統方式的資料傳輸量，以及少的傳輸延遲時間。

Motivated by the promising voice over IP technology, and the wide service availability of WLANs, the application of VOWLAN is expected to experience dramatic growth in the near future. IEEE 802.11e standard was established to guarantee a high level of QoS especially for the time sensitive applications, such as voice services. It introduced a new medium access mechanism HCF in order to solve the QoS provisioning problem in the legacy IEEE 802.11 and its two medium access mechanisms of DCF and PCF. In order to reduce access delays and polling overhead, as well as to enhance the throughput of the channel, we propose an adaptive polling scheme, which works on the HC side in HCCA mode, in which, HC maintains two dynamic polling lists, one contains talking stations and the other contains silence stations, those lists are updated dynamically in order to classify stations into several levels of priority, they are reordered and sorted according to the traffic direction and the queue occupancy on the HC and the station. Stations are classified to be talking or silent according to a simple silence detection mechanism. VBR and CBR codecs are taken into consideration in our silence detection mechanism. TXOPs for stations in the talking list are calculated separately from the TXOPs for stations in silence list. HC then polls stations in both lists in a Round-Robin manner starting from the top of polling list and ending with the tail of silence list. Simulation results show that the polling overhead is reduced significantly, in addition to a high throughout and low access delays comparing to the classical Round-Robin polling scheme and the reference scheme in the standard.