| 摘要: | 在研究中，我們發現台灣地區的降雨模式是與其他預測模型所預測的值不同。在降雨量研究方面，我們從台灣的三個不同降雨量地區四個測站收集12 年內逐時降雨資料。由於一般所能取得逐時降雨量資料太過粗糙,並不能直接反應與信號衰減的關係，國際間亦以逐分降雨率百分之0.01時的值R0.01做為地區降雨特性判斷標準，因此我們採用逐時-逐分的轉換模型。利用中壢地區自行觀測的逐時逐分關係再利用中壢地區12年累積逐時降雨資料以及台灣其他地區逐時降雨資料透過轉換模型可求出台灣地區逐分降雨分佈特徵圖。將得到結果與ITU建議的經驗累積機率密度分佈模型比較，我們發現降雨分佈的 R0.01 比 ITU- R的預測值低。此外，在雨衰減的研究方面，我們利用自行建制的一個點對點的 Ka 波段降雨衰減量測鏈路在中壢地區測量降雨衰減。根據衰減和降雨量的資料，我們亦發現 ITU-R 與Crane模型在降雨衰減預測上一就有高估現象。 在另一方面，我們利用模擬與實驗研究Ka波段電磁波對植被目標物回波的統計特性。假設Ka波段電磁波信號在無法深入植被內部，即可忽視植被中多重散射的現象。利用建立一個半球中隨角度隨機分布旋轉的有限長圓柱體、針狀體，與圓盤來建立樹木的模型，其中，圓柱體用來描述樹幹與樹枝；針狀體則描述針葉；圓盤描述闊葉。 由模擬與實驗結果可知，由接收機接收到的平面波經過植被的散射結果會呈現冪級數分布，依照植被種類分類可知半球隨機分布的針狀葉接收信號呈現Gamma分布，單純樹枝的接收信號則類似lognormal分布且不與旋轉角度有關。 對於圓盤型的闊葉，接收信號則與葉面旋轉角度相關。水平面向的葉子可用類似表面型散射體的Weibull分布描述。而垂直面向的闊葉與混合樹枝與樹葉的種類以及穿透信號則最接近Gamma分布。 最後，我們研究台灣地區使用Ka 波段的 LMDS 系統受降雨的影響程度。利用前述量測的降雨參數，去分析計算LMDS系統效能，誤碼率以及信號干擾比。在多細胞網路服務架構下,小雨反而會提高信號干擾比,因為其他 BTS 干擾在主要 BTS 周遭的干擾信號受到下雨而中斷。 在單一細胞沒有其他細胞干擾影響下QPSK調變可以使用在台灣地區降雨環境。M-QAM調變時則可以服務到四公里範圍。當細胞干擾存在時，BTS 有效服務邊界會依照QPSK，16QAM和64QAM 不同調變的方式分別縮減 2~3 公里和 3~4 的公里甚至無法使用。總之, 若要在多細胞網路環境中使用，將相對應地BTS 有效服務範圍將會被壓縮。為解決干擾與服務半徑問題，我們使細胞規劃擴張到 4 個頻率加 2 個偏極化。然後，比較兩種細胞規劃間M QAM調變的表現，我們發現 4 個頻率的LMDS網路表現比 2 個頻率網路好。因此，在台灣 LMDS 細胞網路環境中，除非使系統拉遠距離或增加頻道，否則M QAM 的調變方式依然無法有效服務至六公里之標準範圍。 In our study, we find that the rainfall pattern in Taiwan area is a little different with other model prediction. In rainfall research, we conducted a research to collect three different rainfall regions in Taiwan from over 10 years hourly rainrate data. Because hourly rainrate data is too rough to react on the relation of attenuation, a conversion model convert hourly to minutely rainfall data is necessary. For establish the rain conversion model to get the long-term R0.01. 1 minute integration rainfall data was collected at one metropolis, Chungli and hourly data with all 4 areas in Taiwan in same period were used. Continuously, we obtain the representable cumulative minutely rainrate distribution function using the 12 years hourly data of Taiwan area we converted then compared with ITU recommended empirical cumulative distribution functions. We find that the R0.01 of rainrate distribution is lower than ITU-R prediction. On the other way, in research of rain attenuation, we build a point to point Ka band terrestrial propagation system with optical rain gauge system to measure the rain attenuation in Chungli. According to the attenuation and rainfall data, we find that ITU-R and Crane models both make overestimates in rainrate and attenuation prediction. In another way, the Ka band signal considered at which the penetration into the canopy is shallow. Thus, the multiple scattering mechanisms can be ignored. A vegetation targets are modeled as a half-space of randomly oriented and spaced finite cylinders, or needles, or disks, or their combinations depending the wavelength. The finite length cylinders stand for tree branch or trunk, the needles for stem or coniferous leaves, and the disks for deciduous leaves. For a plane wave exciting such as canopy simulation and experiment results show that the power distribution scattered or pass through to the receiver from a half-space of needle-shaped leaves follows the Gamma distribution. For disk-shaped leaves, the power returns statistics depend on the leaf orientation. We proved the Weibull distribution provides the best predictions for horizontal oriented leaves, which resemble a surface type scatterer. Gamma distribution is found to best represent the power return from nearly vertical disk-shaped leaves a mixture of branches and leaves and the signal transmit through any vegetation. We also study the rain effects on the performance of a Ka-band LMDS system in Taiwan. The rain distribution statistics is established based on 2-year measurements with 1 minute interval and 12-year measurements with 1 hour interval. The fading due to rain statistics is established based on over 2-year measurements. The system performance in terms of signal-to-interference ratio (S/I), bit error rate, and channel capacity was analyzed. For cellular network service, light rain events induce better S/I because other BTS interference signals around the main BTS are likely to be blocked by rain. The availability of the QPSK modulation scheme in the presence of rain fading without cellular interference meets the margin of BER=10-6 in 6 km cell coverage. When cellular interference presences, the BTS effective service boundary shrinks 2~3 km and 3~4 km using QPSK and 16QAM, respectively, under BER=10-6. Results also suggests that the service radius of a 64QAM modulation scheme with BER=10-6 is less than 4 km without cellular interference but totally not functional under cellular network environment. In this scenario, the 64QAM service radius should be less than 1km and QPSK service radius should be less than 3km according to the channel capacity in rain fading with cellular interference in 6km cell coverage. In summary, in cellular network environment, when we shorten the cell coverage radius, the BTS effective service range will be correspondingly compressed, indicating that the cellular interference is more serious than rain. For solve the interference problem, we expand the cell planning to 4 frequencies and 2 polarizations. Then, the performance of LMDS using M-QAM in 4 frequency network is better than 2 frequency network. In summary, in Taiwan LMDS cellular network environment, M-QAM modulation is difficult to provide an effective and accurate high speed transmission in 6 km large cell coverage radius unless the system expand the frequency channel to increase the distance between each interference sources. |
