An Intelligent eNodeB for LTE Uplink based on Feed Forward Neural Network

Abstract

— In this paper, a feed forward neural network (FFNN) with gradient descent (GD) and Levenberg Marquardt  (LM) training algorithm based framework is used to improve  inter cell interference coordination (ICIC) and radio resource management (RRM)  in LTE system. A neural network based  cognitive engine is embedded within eNodeB which coordinately suggest optimal radio parameters to the users, and best transmit power to the operating users by neighboring cells. Long term learning, fast decision making , and less computational complexity are the three main requirements to map CE to distribute systematically in any cognitive communication system and most of the present techniques used as a cognitive solution lack in. The mechanism of feed forward network supported framework is examined with traditional schemes. To ensure a better performance of the system, the results are verified and compared with traditional schemes.