30 April 2003 in Microwave and Optical Technology Letters
In this paper, the genetic algorithm (GA) has been applied to calculate the optimized length and width of rectangular microstrip antennas. The inputs to the problem are the desired resonant frequency, dielectric constant, and thickness of the substrate; the outputs are the optimized length and width. The antennas considered are electrically thin. Method of moments (MoM)‐based IE3D software from Zealand Inc., USA, and experimental results are used to validate the GA‐based code. The results are in good agreement. © 2003 Wiley Periodicals, Inc. Microwave Opt Technol Lett 37: 431–433, 2003; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/mop.10940
10 October 2011 in International Journal of Parallel, Emergent and Distributed Systems
SGSastry V.R.S. Gollapudi
+1
This paper proposes a new variant of parallel particle swarm optimisation (PPSO) known as small population-based modified PPSO (SPMPPSO) for fast motion estimation. The proposed technique is used to reduce computational time for block motion estimation in video. In the said technique, the velocity and position equations of PPSO are modified to achieve adaptive step size for getting true motion vector. The new position of swarm depends on previous motion vector, time decreasing inertia weight and on time-varying acceleration coefficient. The best matching block is predicted by step size/position equation of SPMPPSO. The Von Neumann topology is used as search pattern in the SPMPPSO. In SPMPPSO, small population, i.e. five swarms with which two-step search are used to find best matching block. Zero motion prejudgement is used leads to faster convergence for getting the motion vector. The results of SPMPPSO are compared with those of PPSO and with those of other motion estimation algorithms. The limitations such as computational time, search parameter, initial search and search space are overcome in SPMPPSO. The proposed technique saves computational time up to 94% when compared with other published methods.
1 January 2010 in IETE Journal of Research
SGSastryV. R. S Gollapudi
The paper presents an improved variant of bacterial foraging optimization (BFO) named as Synchronous Bacterial Foraging Optimization (SBFO). The proposed SBFO is used to optimize multimodal and high dimensional functions. As all the bacteria update their information simultaneously, it has been named synchronous. A mutation operator proposed in this paper performs the global search. In SBFO, all bacteria process information independently in the same generation; hence parallel computers can be used to evaluate fitness values. The performance of SBFO is validated on a set of seven benchmark functions i.e. Sphere, Rosenbrock, Rastrigin, Griewank, Ackley, Schaffer's, Shekel's Foxholes. The results are compared with other published methods such as classical BFO, hybrid BFO (BSO), swarm-based algorithms, differential evaluation and Ashaker. The simulation results on benchmark functions show that the proposed optimization is capable of producing good quality global optima as compared to the above mentioned optimization techniques. The proposed method is a new variant of BFO and suitable for optimization of multimodal and high dimensional functions.
23 August 2002 in Microwave and Optical Technology Letters