Rita Mahajan

Implementation of Single Artificial Neuron Using Various Activation Functions and XOR Gate on FPGA Chip

In this paper, a hardware implementation of an artificial neural network on Field Programmable Gate Arrays (FPGA) is done step by step. First single neurons are implemented using different activation function and then a XOR Gate is implemented. The concurrent structure of a neural network makes it very fast for the computation of certain tasks. This makes ANN well suited for implementation in VLSI technology. Hardware realization of a Neural Network depends on the efficient implementation of Artificial single neuron. FPGA-based reconfigurable computing architectures are suitable for hardware implementation of neural networks. FPGA realization of ANNs with a large number of neurons is still a challenging task. Here design and implementation of a Single neuron with various activation functions is done and then an XOR Gate is implemented using one of these designs. The design of XOR Gate is done by using gradient descent algorithm and taking 3-layers 5-neurons model.

Performance analysis of cyclostationary and energy detection spectrum sensing techniques

The current policies of spectrum allocation leads to inefficient use of spectrum, making it scarce but the truth is, spectrum is not scarce but is underutilized. In todays era dynamic spectrum access is the need of the hour and cognitive radio is a novel technology which improves spectrum utilization. In this paper overview of cognitive radio along with spectrum sensing techniques is taken into account. Energy detection based spectrum sensing techniques and cyclostationary detection based spectrum sensing techniques are discussed in detail with their block diagrams and flow charts. The paper also describes the performance analysis of energy detector over various fading environments viz. AWGN, Rayleigh, Rican and MIMO. The energy and cyclostationary detectors are implemented in matlab and their simulation results are presented using graphical analysis with the help of ROC (receiver operating characteristics) curves.