Performance of OOK with ANN Equalization in Indoor Optical Wireless Communication System

Abstract: The artificial neural network (ANN) could be a valuable signal processing tool for future communication systems as it has some attractive properties like adaptability, parallel processing, and universal approximation. ANN has been purposed as an alternative to traditional finite impulse response digital filter for equalization for communication and in many cases ANN outperforms the FIR equalizer especially in non-linear channel. A well train ANN can perform the task equalization efficiently. Most of the ANN application as an equalizer is focused on RF domain. In this paper we use the ANN for equalization in indoor environment in an optical channel. The simulation results show the performance of ANN is as good as FIR filters equalizer. I. INTRODUCTION: Performance of a communication systems can be enhanced by using the ANN as a tool for signal detection and equalization [1-5]. ANN based signal detection has been suggested for non-stationary and non-linear channel because of its adaptability and ability to classify the nonlinear problem. The signal detection in non-stationary environment can be reformulated as adaptive pattern classification problem and neural network can be applied for classification [1]. The RF bandwidth is already congested because of high bandwidth applications such as the internet, video conferencing etc. Thus the need for an alternative means of communication to address the bandwidth congestion problems. One technology capable of addressing this problem is the optical communications (fibre or free space) particularly if high data rate is the main priority as it provides unlimited bandwidth. Optical wireless communication system (OWS) is already adopted in modern camera phones, digital cameras and other devices (500 millionth infrared (IR) transceiver been sold from 1994 to 2005 [6]). OWS can be used both indoor and outdoor as a line-of-sight (LOS) or non-LOS. For outdoor the LOS with tracking is the most widely used [7], whereas the non-LOS is employed for indoor applications. Diffused (non-LOS) links suffer from multipath induced intersymbol interference (ISI), thus resulting in reduced bandwidth. Equalization techniques have been used to minimize the effects of ISI [10]. The traditional equalizers are generally finite impulse response (FIR) digital filter, which are principally based on the problem of finding the inverse filter of the channel. However, it is not always possible to get the inverse filter, if channel has zero at origin, the inverse filter fails to exits. In such cases, ANN based equalizer has been purposed and studied [3, 4]. Different modulation techniques have been purposed for indoor OWS including on-off keying (OOK), pulse position modulation (PPM), pulse interval modulation (PIM), and dual-header pulse interval modulation (DH-PIM) [8-10]. OOK format is the most basic and widely studied scheme for indoor OWS with intensity modulation and direct detection (IM/DD). Its popularity is mainly due to simplicity in transceiver design and implementation. Though the power requirement is much higher than other modulation schemes it is less affected by the multipath induced ISI [11]. Most of the ANN based equalizations are studied in RF domain. Wavelet–ANN based equalization and detection for OWS is purposed in [12, 13] in which the authors have reformulated the signal detection as a problem of feature extraction and pattern reorganization. In this paper, the performance of ANN based OOK for indoor OWS is investigated. The simulation results show the performances of ANN based equalizer is identical to the traditional FIR filter equalizer. The rest of the paper is organized as follow. Section 2 gives brief review of channel model and noise model of indoor WOS, followed by a section on adaptive equalizer and ANN. The purposed model is described in section 4 and the simulation results and concluding remarks are given in sections 5 and 6, respectively.