Two modern optimization methods including Particle Swarm Optimization and Differential Evolution are compared on twelve constrained nonlinear test functions. Generally, the results show that Differential Evolution is better than Particle... more
Two modern optimization methods including Particle Swarm Optimization and Differential Evolution are compared on twelve constrained nonlinear test functions. Generally, the results show that Differential Evolution is better than Particle Swarm Optimization in terms of high-quality solutions, running time and robustness.
Sidelobe level suppression is a typical circular array (CA) optimization problem. Many conventional techniques are proposed in the due course of time which are computationally complexed and time consuming. As an alternative, several... more
Sidelobe level suppression is a typical circular array (CA) optimization problem. Many conventional techniques are proposed in the due course of time which are computationally complexed and time consuming. As an alternative, several evolutionary and heuristics approaches have evolved for solving such problems. In this paper, a novel accelerated particle swarm optimization (APSO) is considered for solving circular array optimization problem. The APSO is implemented for CA synthesis with the objective of SLL suppression. The objective is achieved under scanned and unscanned beam conditions. A very low SLL of -25dB is observed when the main beam positioned at 00,150 and 300 . In all the cases the efficiency of the APSO is evaluated by the comparing the generated radiation patterns with those obtained from uniform CA.
In this paper, an optimized Recursive One-Sided Hypothesis Testing (ROHT) threshold estimation algorithm for energy detection based on Cognitive Radio (CR) application is presented. The ROHT algorithm is well known to compute and correct... more
In this paper, an optimized Recursive One-Sided Hypothesis Testing (ROHT) threshold estimation algorithm for energy detection based on Cognitive Radio (CR) application is presented. The ROHT algorithm is well known to compute and correct threshold values based on the choice of the parameter values; namely the coefficient of standard deviation (z-value) and the stopping criteria (). A fixed computational process has been employed in most cases to estimate these parameter values, thus rendering them non-adaptive under different sensing conditions. Also, this fixed (manual tuning) process requires prior knowledge of some noise level to enable pre-configuration of a predefined target false alarm rate. This renders the parameter estimation process cumbrous and unworkable for real-time purposes, particularly for autonomous CR applications. Furthermore, using wrong parameter values may lead to either too high or too low false alarms or detection rates of the algorithm. Sequel to aforementioned mentioned constraints, we propose a new mechanism for instantaneous parameter optimization of the ROHT algorithm using Particle Swarm Optimization (PSO) algorithm. Our PSO-ROHT model design was extensively tested under different conditions and results were compared to the non-optimized ROHT. The results obtained show that the proposed design effectively adapts the parameter values of the Recursive One-Sided Hypothesis Testing algorithm in accordance with the input dataset under consideration. Also, that the proposed optimized model outperforms its non-optimized counterpart following the estimated detection probability and false alarm probability of both schemes, particularly in detecting Orthogonal Frequency-Division Multiplexing signals. In detecting the Orthogonal Frequency-Division Multiplexing signals at signal-to-noise ratio of 3dB and above, the proposed model achieved a higher detection rate of 96.23% while maintaining a low false alarm rate below 10%, which complies with the IEEE 802.22 standard for Cognitive Radio application. Our PSO-ROHT algorithm is shown to be highly effective for autonomous and full blind signal detection in CR, with strong potentials for application in other areas requiring automatic threshold estimation.
Two modern optimization methods including Particle Swarm Optimization and Differential Evolution are compared on twelve constrained nonlinear test functions. Generally, the results show that Differential Evolution is better than Particle... more
Two modern optimization methods including Particle Swarm Optimization and Differential Evolution are compared on twelve constrained nonlinear test functions. Generally, the results show that Differential Evolution is better than Particle Swarm Optimization in terms of high-quality solutions, running time and robustness.
