捕捉與解決多重跳接式無線網路中的干擾

Abstract

在多重跳接式無線網路中沒有精確的網路部署規劃，干擾仍會發生在裝置密集分佈的區域，無法完全靠頻率重複使用避免裝置間干擾，而且逐跳轉送資料可能會產生更多同時存在的網路流量，造成更嚴重的裝置間干擾。另一方面，考慮到橋接異質無線網路的裝置，使用鄰近通道的不同無線接取技術可能會造成裝置內的干擾。因為異質的收發器位於同一個裝置之中，某一個收發器可能會遭受來自其他收發器，並且幾乎沒有能量損失的干擾。因此，不只是裝置之間的干擾，還有裝置之中的干擾都可能會造成相當可觀的效能降級。 本研究致力於設計與開發一個多重跳接式無線網路系統，並驗證其功能以及評估該系統之效能，特別是干擾造成的衝擊。實驗結果顯示多重跳接式無線接取網路系統遭受由控制面向中的廣播行為、資料面向中的轉送行為所產生之顯著的負擔。為了估算在複雜部署情境中的裝置之間的干擾，本研究捕捉了在真實環境中的封包紀錄和環境影響，並且在實驗室中同步地重製它們。實驗結果顯示由封包所組成之事件的事件重製率達到了95%。另一方面，考慮到異質網路裝置，因為共存所產生的裝置內干擾所造成的衝擊也被檢視與量化。裝置內干擾導致為了維持可接受的吞吐量所需求的訊號能量等級提高。測量結果顯示裝置間與裝置內干擾的影響在訊號能量等級上有9dB的差距。 為了解決干擾的議題，本研究提出了考慮流量特性的多通道切換機制。被提出的察覺流量的切換方法使用較少的介面利用多通道，並且使通道切換行為符合網路流量負載。實作在商用產品上的察覺流量的切換方法，證明了在雙通道單介面的設定，相對於單通道單介面設定有75%的吞吐量改進。此外，被提出的察覺流量的分群方法將平行流量分散到不同通道中以提升吞吐量，以及聚集相關的流量到相同通道中，以降低通道切換負擔與提升通道利用率。模擬的結果顯示，相較於基於拓樸的解決方案，察覺流量的分群方法有38%的總吞吐量改進。

Devices in wireless multi-hop networks (WMNs) are deployed without precise network planning. Interference among devices cannot be avoided completely by frequency reuse and may occur in densely distributed regions. Forwarding data hop-by-hop may result in concurrent traffic flows and introduce significant inter-device interference. On the other hand, considering a device which bridges heterogeneous wireless networks, forwarding data between different wireless access technologies using adjacent channels may cause intra-device interference. Because heterogeneous transceivers are within a co-located device, a transceiver may suffer the interference from other transceivers with little loss of power. Therefore, perceptible performance degradation may result from not only interference among devices but also interference inside a device. This work is devoted to designing and developing a WMN system, verifying and validating the functionalities, and evaluating the system performance, especially impacts of interference. The experimental results reveal that the WMN system suffers notable overheads caused by the interference during the control-plane broadcasting and the data-plane forwarding. To assess inter-device interference in complicated deployment scenarios, we capture packet traces and environment effects in the real environment and reproduce them synchronously in the lab. In the experimental results, the reproduction ratio of the events consisting of packets achieves 95%. On the other hand, considering heterogeneous wireless networks, impacts of intra-device interference are also examined and quantified for their co-existence. Intra-device interference raises the required power level to maintain acceptable throughput. The evaluation results exhibit that the influences of intra-device and inter-device interference have 9dB difference of power levels. To resolve the interference issues dynamically, we propose multi-channel switching schemes with knowledge of traffic patterns. The proposed traffic-aware switching scheme (TRASS) utilizes multiple channels with fewer radios and adapts channel switching to traffic loads. TRASS implemented on the commercial device demonstrates 75% throughput improvement by the 2-channel 1-radio configuration over the 1-channel 1-radio configuration. Moreover, the proposed traffic-aware clustering scheme (TCS) separates parallel traffic flows into different channels to enhance throughput, and gathers correlative traffic flows in the same channel to reduce switching overheads and raise channel utilization. The simulation results show that TCS achieves 38% improvement of aggregate throughput over the topology-based solution.