Adnan Quadri & Dr. Naima Kaabouch Optimization Efficiency Particle Swarm Optimization Based Adaptive Filter for Noise Cancellation in Cognitive Radio Systems Adnan Quadri & Dr. Naima Kaabouch Department of Electrical Engineering, College of Engineering & Mines, University of North Dakota, USA Introduction Methodology Results Conclusions In wireless communications, signals are corrupted by noise, interference, path loss, and fading. In traditional communication systems, fixed hardware based filters are employed for noise cancellation but they are costly, bulky, and can filter only specific frequencies. However, next generation communication systems, such as Cognitive Radios (CR), will be reconfigurable and can dynamically adjust their parameters to filter any signal of any frequency. GOAL AND OBJECTIVES The goal of this project is to develop efficient reconfigurable filtering algorithms that meet the requirements of the next generation communication systems. To achieve this goal, the following objectives are pursued: Investigate evolutionary algorithms for denoising signals in cognitive radio systems. Compare the efficiencies of evolutionary algorithms with those of the existing techniques. Implement the techniques using Software Defined Radio (SDR) and perform real experiments. At the transmitter end, generated bit stream is modulated using BPSK modulation and transmitted as signal 𝑥(𝑡). At the receiver end, additive white Gaussian noise, 𝑛(𝑡) is added to the received signal, 𝑟(𝑡). Received noisy signal is sampled and forwarded to the Adaptive Noise Cancellation block. The adaptive filtering system is based on the system design of an adaptive line enhancer (ALE). PSO is initialized for a defined particle size and iteration, during which PSO searches for optimum solution to minimize the filter error signal, 𝑒[𝑛]. Flowchart for PSO algorithm Performance Evaluation Metrics: Bit Error Rate (BER) is defined as: ( 𝑅 𝑏 − 𝑇 𝑏 )/ 𝑁 Where, N = Length of received signal, 𝑅_(𝑏 )= Number of received bits, 𝑇_(𝑏 ) = Number of Transmitted bits. Mean Square Error (MSE) is defined as: 1/𝑁 𝑙=1 𝑁 (𝑁𝑜𝑖𝑠𝑦 𝑆𝑖𝑔𝑛𝑎𝑙 −𝐹𝑖𝑙𝑡𝑒𝑟 𝑂𝑢𝑡𝑝𝑢𝑡) 2 PSO algorithm was implemented for adaptive noise cancellation in Cognitive Radio systems. Extensive simulations were performed to evaluate and compare the performance of PSO with that of LMS. Results show that PSO performs significantly better as both BER and MSE of PSO are lower than the BER and MSE of LMS under varying SNR conditions. Future work includes developing the Genetic Algorithm based denoising technique, implementing all algorithms using Software Defined Radios (SDR), and perform real experiments. BER of PSO and LMS for noisy signal over varying Signal-to-Noise Ratio (SNR) Goal and Objectives MSE of PSO and LMS over varying SNR conditions References A. Quadri, M. Riahi Manesh, & N. Kaabouch, “Denoising Signals in Cognitive Radio Systems Using An Evolutionary Algorithm Based Adaptive Filter”, IEEE Annual Ubiquitous Computing, Electronics & Mobile Communication Conference, pp. 1–7, 2016. Impact of particle size on PSO convergence characteristics Yes No A. Quadri, M. Riahi Manesh, & N. Kaabouch, “Noise Cancellation in Cognitive Radio Systems: A Performance Comparison of Evolutionary Algorithms”, IEEE Annual Computing and Communication Workshop and Conference, pp. 1–7, 2017. Methodology Mean square error for different LMS step sizes We implemented a Particle Swarm Optimization (PSO) algorithm. This algorithm is one of the evolutionary algorithms that is based on stochastic global optimization technique. PSO is inspired by the social interaction of bird flocking. Birds in search of food update the best position of food source among the group. Similarly, PSO minimizes the noise in a signal by searching for an optimal solution, which in the case of adaptive filters are weight coefficients. Table: Comparison of PSO and LMS performance Algorithm Complexity Convergence Optimization Efficiency PSO Complex Not affected by initialization variables Able to perform global optimization LMS Simple Affected by initialization variables, e.g. step size Only performs local optimization M. Riahi Manesh, A. Quadri & N. Kaabouch, “An Optimized SNR Estimation Technique Using Particle Swarm Optimization Algorithm”, IEEE Annual Computing and Communication Workshop and Conference, pp. 1–7, 2017.