Mediterranean Conference on Embedded Computing
,,/"
Bar, Montenegro
MECD 2012 -
An Intelligent Approach to Link Parameter Estimation for Outdoor Optical Wireless Channels M.Ali
A. EI Yakzan, R.J. Green, E.L. Hines
School of Engineering University of Warwick Coventry, CV4 7AL, United Kingdom
[email protected]
College of Computer Science and Engineering University of Ha'il P.O. Box 2440, Ha'iI8 1451, Kingdom of Saudi Arabia
[email protected]
Abstract- An intelligent approach to establishing link parameters of importance in the optical wireless channel is proposed. A genetic algorithm (GA) is used to perform a selection technique for the link parameters with minimal cost. An artificial neural (ANN) network then adapts the channel with the best fit of learning sets. The objectives include
an
attempt
to
exploit
the
power
of
hybrid
intelligent systems in the outdoor optical wireless channel and a prediction of a reliable adaptation technique that contributes to best possible bit-error-rate
(BER)
under
certain weather conditions and link characteristics.
However, the transmission process imposes many challenges in the outdoor environment that threaten its survivability, such as due to adverse weather conditions, vibrations of buildings, eye safety requirements, passing obstacles and multipath effects in the indoor environment. Several approaches, as discussed in the literature have been analyzed in order to propose ways of treating these challenges. In this paper, we try to show different weather models designed for analyzing the optical wireless performance in order to stress on external channel parameters that could be controlled.
Keywords: BER; genetic algorithm; artificial neural network.
I.
INTRODUCTION
Optical wireless is the latest approach in telecommunication media. It refers to the integrated use of two technologies: the radio frequency (RF) and the optical fiber. These waves and optical fiber technology have the potential of providing data capacity of the order of Gbps and Tbps respectively. The requirements of broadband wireless system can be met through the integration of the optical fiber with millimeter wave wireless systems - which is the subject of the latest area of telecommunications research, investigating both the reliability and the confidentiality of multimedia services. Optical Wireless, based on free-space optics (FSO), is becoming an attractive alternative for indoor and outdoor communications because of its extremely wide bandwidth, high speed, low cost, license free spectrum and interference security. The direct interception of an optical wireless beam between the two remote networking locations is basically impossible because any individual attempts to block the beam would occur near the optical wireless equipment terminus point. Consequently, it is not necessary in the meantime to add security protocols, especially when higher layer protocols (transport layer) can be used in conjunction with layer one optical wireless technology to encrypt sensitive network information and to provide additional encapsulation.
Figure
1.
Optical Wireless Channel Model.
Due to this analysis, relations among parameters will be discussed, drawing attention to parameters of major importance to system performance. Starting with the choice of the optical transmitter, through the optical channel and then to optical to electrical conversion at the receiver, many parameters are directly related to each other, which together decide about the system performance. In industry, designers setup the received power needed to obtain an acceptable bit error rate and anything beyond this level is not accomplished. This paper aims to put everything together with an open discussion relating to the parameters together even if the received power falls below the threshold. An attempt to improve the link margin is a key major factor, even if it does not directly lead to an accepted reception. Therefore, the cost of external parameters will be selected as the minimum; for instance, no more than enough power is transmitted, not more than needed bandwidth is considered. Moreover, different modulation techniques are discussed, to analyze the last possible step of improvement that could be accomplished
Mediterranean Conference on Embedded Computing
,,/"
when analyzing the signal to noise ratio or the bit error rate at the receiver. Section Two discuses some pertinent literature about the achievements of intelligent systems in the optical wireless field. Section Three shows the variables of link importance and different controllable parameters in the optical wireless channel. Section Four discusses the different weather models implemented so far. This part will also take two considerations into account: the channel with and without the effect of turbulence. The methodology of the proposed work is discussed further in Section Four, where it shows a new approach to an intelligent selection technique of parameters after studying the link conditions. Section Five discusses the upcoming work and other achievements. [I.
Bar, Montenegro
MECD 2012
PREVIOUS ACHIEVEMENTS OF [NTELLIGENT SYSTEMS IN OPTICAL WIRELESS SYSTEMS
In the intelligent system approach, a genetic algorithm controlled multiport transmitter is proposed in [ 1]. It describes an alternative approach of power optimization in mobile indoors optical wireless systems. The algorithm stresses on dynamically changing the intensity of individual diffusion spots in multiple rooms, showing a negligible impact on bandwidth and RMS delay spread, to emphasize the use of more effective transmitter and receiver in various scenarios and application. The simulations show that the dynamic range of the rooms, referenced against the peak received power, can be reduced by up to 26% when empty and to 12% when user movement obstructs the channel. Moreover, a genetic algorithm for improving the received SNR in multi-spot diffuse infrared communication systems is proposed in [2]. [t provides an improvement of up to 8 dB compared to LSMS at a cost of about 1% overhead by adjusting the position and number of the spots on the ceiling as the receiver moves to different locations. An adaptive hybrid modulation scheme is proposed in [3] by considering the three major modulation schemes in the optical wireless channel. On-off keying (OOK), pulse amplitude modulation (PAM) and pulse position modulation (PPM) are analyzed in terms of their power and bandwidth requirements. The proposed modulation scheme takes the real-time channel conditions into account, which is different from other schemes. By adaptively employing amplitude and position modulation, a guaranteed system performance can be secured without compromising power and bandwidth efficiency. Channel Optimization for a PPM system is achieved in [4] by training a Multilayer Perceptron (MLP) artificial neural network for decision coding in different bit resolution and spread delays of the optical channels. The study shows that for a highly diffuse channel, it is necessary to have a 10 dB SNR to achieve an acceptable bit error rate of 10.5• ANN based equalizer is [0 dB lower compared to the unequalized soft decoding at 155 Mbps data rate. The results indicate that for all ranges of delay spread, neural network equalization is an effective tool of mitigating the lSI.
-
The type of modulation in the optical wireless link also affects the type of optimization done on several parameters, which has been discussed in [5]. Actually, multicarrier modulation permits transmission with minimal inter-symbol interference on selective frequency channels. Electrical MCM use hundreds of carriers for modulations, in the channels, there is a drawback of high peak-to-average power ratio (PAR) that can lead to clipping and nonlinear distortions; therefore, QPSK and BPSK techniques are used to minimize this effect by selecting specific harmonics for signal recovery. Optical wireless systems involve modulation of digital and analog information from different electrical subcarriers which are then modulated onto a single optical carrier. The multi-subcarrier system can be modulated using FM, PM and Intensity Modulation (1M) [6]. The latter, along with direct detection (DD) is being used for most MSM systems, which permits asynchronous multiplexing of heterogeneous information and permits the receiver to make selective demodulation only for the streams of interest. For this advantage, MSM of [M/DD is widely used in optical transmissions of different data sources [7]. Basically, average power reduction techniques for intensity-modulated optical systems have been approached using multiple BPSK and QPSK subcarriers. One technique involves block coding of the data bits to be transmitted and the amplitude of symbols modulated on the subcarrier, the second replaces the fixed DC bias by a variable one that changes on a symbol-to-symbol basis. These techniques are applied to subcarriers originating from a single transmitter with symbol-synchronization [8]. A genetic algorithm based selection of a transmission wavelength in the LOS optical wireless channel is presented in [9]. It shows a selection technique of appropriate transmission wavelengths in the line-of sight (LOS) optical wireless channel under different weather conditions. The paper revealed a 10 dB/km decrease of attenuation using a GA based selection of wavelength when adverse weather conditions are deployed. This achievement shed light on the maximum link margin that could be controlled by adapting the input parameters to the channel characteristics to maintain cost effective usage of parameters with acceptable signal-to-noise performance. III.
VARIABLES OF POSSIBLE LINK IMPORTANCE
1.
Transmitter power in Watts, optical power, or optical power density in Watts per square meter emerging from the transmitter optics.
2.
Transmitter beam divergence, in degrees, a parallel beam would have 0° beam divergence, assuming no scattering.
3.
Transmitter optical wavelengths if a LED transmitter, or wavelength if a laser.
4.
Modulation type (digital/analogue).
Mediterranean Conference on Embedded Computing
,,/"
Bar, Montenegro
MECD 2012
5.
Modulation bandwidth.
6.
Low level transmitted optical power, related to extinction ratio for a digital system, relative to high level optical power as a proportion. For example, 1 mW to 100 mW would be an extinction ratio of 100.
7.
Environmental attenuation due to fog, mist, rain, snow, sandstorm, etc.
8.
Environmental scattering leading to another attenuation term like scintillation.
9. Spatial variations in optical link attenuation over the distance between the transmitter and receiver. This is due to the change of refractive index of the electromagnetic wave medium. 10. Time variations in optical link attenuation and this is related to a probability model that estimates the fading intensity according to the scintillation index.
-
IV.
METHODOLOGY
The optical wireless layer in the short range LOS channel is subject to time-variant environmental challenges. There are several internal and external factors that determine the overall channel performance. Therefore, many levels of optimization have been approached so far. In this paper, we propose single and multi-objective selection techniques based on OaK modulation for the purpose of data survivability in the optical wireless channel, taking into consideration many variables like: average transmission power/laser, number of laser, transmitter beam divergence, diameters of transmit and receive apertures, the receiver sensitivity, deployment distance, the deterministic weather, bandwidth of the signal, its wavelength, and the bit rate. The parameters will be generated for a large number of combinations covering the whole ranges, some of which will be chosen to be fixed for the first stage approaches. The flow chart of the system appears in Figure 2.
1 1. Receiver optical aperture from which the received power can be calculated by multiplying the area of the lens or optical structure by the power per unit area arriving at the receiver optics. 12. The responsivity of the photodetector In converting optical power to electrical current. 13. The ambient optical received power due to daylight, starlight and moonlight.
No
Adjust Weights �====C: Yeo
14. The shot noise in the receiver due to the ambient received optical signal. 15.
The shot noise in the receIver due to the received optical power which has been sent from the transmitter through the environment to the receiver and converted into a detected signal current in the receiver's photodiode.
16. The receiver electrical signal bandwidth. 17.
The receiver noise bandwidth, normally similar to the electrical signal bandwidth, but not normally exactly the same.
18. The signal to noise performance of the front end of the receiver-amplifier, based on the actual design. For example, high impedance frontend, transimpedance frontend, photoparametric amplifier/frequency changer. This is important to know how much shot noise and/or thermal noise (Johnson noise) is generated in the frontend amplifying device and thus, a FET or bipolar transistor.
Figure
2. Flowchart of Proposed System.
Genetic algorithms belong to the class of evolutionary algorithms (EA), which generate solutions to optimization problems using techniques inspired by natural evolution, such as inheritance, mutation, selection and crossover. In such intelligence, a population of strings (like chromosomes) encodes candidate solutions to an optimization problem. In each stage, stochastic selection of pairs is done according to Darwin's survival of the fittest, then modified and sometimes mutated. After a large number of iterations, the fittest candidates tend to converge to the optimal value if the algorithm fmds a way to the solution (Figure 3).
Mediterranean Conference on Embedded Computing
,,/"
Bar, Montenegro
MECD - 2012
I"pul- OUlplll HelatioftS "'or the be t parameler
DetermInIstic Attenuations
Atm. Atte uatio n Fog
