基於類神經網路在感知網路下進行頻譜偵測

感知網路中的重要技術頻譜感測(Spectrum Sensing)，主要目的是次要使用者藉由信號來判斷主要使用者存在與否。為了提升頻譜感測的效能，現在都是採用合作式的頻譜感測。合作式頻譜感測的重要關鍵融合定律(Fusing Rule)，用途是降低頻譜間使用者發生干擾的機率。傳統的融合定律可以分類如下: OR Rule，AND Rule，N out of K Rule，Majority Rule以及Optimal Rule，上述的融合定律都需要兩項參數，SNR和能量門閥值(Threshold)，來進行融合法則的運算和判斷。然而感知網路和主要網路是兩個不同的系統，所以這兩項數值不容易量測的到。在本篇論文中，使用一個智能網路模型---類神經網路，將訓練後的類神經網路當成頻譜感測的融合中心(Fusing Center)。假設在一個SNR已知的類神經網路環境，提供越多的網路訓練資料，類神經網路表現的錯誤率和Optimal Rule結果相近似甚至更好。當SNR無法得知的環境下，類神經網路在訓練之後的效能也能接近已知SNR環境下的效能。

關鍵字

感知網路；合作式頻譜感測；融合規則；類神經網路

並列摘要

Spectrum sensing is one of the important technologies in a Cognitive network, and the main purpose for spectrum sensing is detecting primary users are present or not for secondary users by signal. In order to promote the number of correct result, Cooperative Spectrum Sensing has been proposed to solve this predicament. Fusing Rule is the key point in a Cooperative Spectrum Sensing, and is using for avoiding users interference. Traditional Fusing rules can be classified as follows, OR rule, AND Rule, Majority Rule, N out of K Rule, and Optimal Rule. All the Fusing Rules in above needed two elements, SNR and Threshold, to make a decision. However, there are different network system between Cognitive network and primary network, therefore SNR and threshold is difficult to measure. A Neural Network has been proposed for the Cognitive network Fusing center, in SNR known condition, the Neural network error probabilities are similar and even better than Optimal Rule when a large number training data is informed. When facing with an SNR unknown environment, Neural Network error probability performance is similar to the result in an SNR known environment.

並列關鍵字

Cognitive Network ； Cooperative Spectrum Sensing ； Fusion Rule ； Neural Network

參考文獻

[ 1 ] . R. W. Thomas, L.A. DaSilva, A.B. Mackenzie, “Cognitive Networks,” in Proceedingsof the First IEEE International Symposium on New Frontiers in Dynamic SpectrumAccess Networks, Baltimore, MD, USA, November 8-11, 2005.

Google Scholar

[ 2 ] . Federal Communications Commission, “Spectrum Policy Task Force,” Rep. ETDocket no. 02-135, Nov. 2002.

Google Scholar

[ 3 ] . Ghasemi, E. S. Sousa, “Spectrum Sensing in Cognitive Radio Networks:Requirements, Challenges and Design Trade-offs,” IEEE Communications Magazine,vol. 46, no. 4, pp. 32-39, Apr. 2008.

Google Scholar

[ 4 ] . Kang G. Shin, Hyoil Kim, Alexander W. Min, Ashwini Kumar, University of Michigan, “Cognitive Radio For Dynamic Spectrum Access: From Concept To Reality”IEEE Communications, pp. 64-74, Dec . 2010.