CLIPPING DISTORTION IN OPTICAL WIRELESS COMMUNICATION SYSTEMS: MODEL, SIMULATIONS AND STATISTICS Omar Falou and Xavier N. Fernando (ofalou, fernando@ee.ryerson.ca) Dept. of Electrical and Computer Engineering, Ryerson University, Toronto, Canada ABSTRACT Clipping distortion has been a subject of interest in many recent studies on frequency-division subcarrier multiplexing (SCM) used in cable television applications and wireless communication systems. Clipping occurs when the modulating signal current, which drives the laser diode, occasionally drops below the laser threshold current, turning the laser off. A model was built to investigate this phenomenon. A Rayleigh fading channel was used to simulate the wireless channel. It was found that, for channels with four fading paths, an increase in the number of these paths for the same fading channel will not increase the clipping percentage by as much as it was expect. Furthermore, an increase in the threshold current of the laser diode can significantly minimize the signal distortion. 1. INTRODUCTION Communications through optical fiber have become a major information-transmission system due to many advantages it provides. Networks based on optical fiber technology provides, fast, efficient and reliable means of transmitting information compared to other communication systems such as wireline and wireless systems [1]. A fiber optic transmission system consists of three basic elements: the optical transmitter, the fiber optic cable and the optical receiver. The optical transmitter converts electrical signal into an optical signal. The source of the optical signal can be either a Light Emitting Diode (LED) or a Laser Diode (LD). LDs have some advantages over LEDs since they can be focused to a very narrow range, allowing control over the angle of incidence. They also preserve the character of the signal over long distances, which make them ideal for long transmission distances [2].However, one of the disadvantages of LDs, is that signal distortion can occur, mainly the clipping distortion. Fig. 1. Relationship between optical output power and laser diode drive current [1] Clipping distortion has been a subject of interest in many recent studies on frequency-division subcarrier multiplexing (SCM) used in cable television (CATV) applications [3]-[5]. Clipping occurs when the modulating signal current, which drives the laser diode, occasionally drops below the laser threshold current, turning the laser off [3],[6]. In this project, we build a Simulink model to simulate an optical transmission system having a LD as its optical transmitter. Then, we use this model to investigate the effect of various Rayleigh fading channels on clipping distortion. 2. OVERVIEW OF CLIPPING DISTORTION For laser diodes, the relationship between optical output power and diode drive current is shown in Fig. 1. At lower drive currents, only spontaneous radiation is emitted, therefore, no increase in the power output is observed (Region 1). As the drive current reaches the lasing threshold, a sharply increase in optical power output occurs (Region 2). This region is called stimulated emission region or linear region. At high power outputs, the slope of this curve decreases due to junction heating and saturation is reached (Region 3) [1]. As depicted in Fig. 1, the laser exhibits a threshold current, Ith, and the drive current is biased to a DC Fig. 3. Small-scale fading manifestations and degradations Fig. 2. Fiber optic transmission system used to study clipping distortion operating point, IB [7]. Clipping occurs at the zero power level when the drive current falls below the threshold current Ith. In order to study clipping distortion, let us consider the following optical transmission system shown in Fig. 2. Initially, the signal s(t) is modulated using Quadrature Phase Shift Keying (QPSK). Then, the resulted analog signal is passed through a Rayleigh fading channel in order to simulate a wireless communication channel. After that, the modulated signal is transformed into optical signal or power using the Laser Diode. The optical signal is transmitted in the optical fiber where it’s detected by the photo detector, converted into current, then demodulated to regenerate the original signal s(t). The clipping distortion happens while the signal x(t) is being converted into optical power P(t). It occurs when the intensity of the injected current x(t) is less than the threshold Ith as mentioned above. A key component of the fiber optic transmission system is the Rayleigh fading channel, which gives rise to amplitude fluctuations which are the main cause of clipping. This channel will be studied in details in the next section. 2. RAYLEIGH FADING CHANNEL In a wireless communication system, a signal can travel from a transmitter to a receiver over multiple reflective paths. This phenomenon is called multipath propagation. This can cause fluctuations in the amplitude, phase or angle of arrival, of the transmitted signal at the receiver’s end, which gives rise to what we call multipath fading [8]. Fading can also be caused by channel variation due to motion between transmitter and receiver, physical changes in the medium and shadowing effect of large obstructions [8]. The variations of amplitude, phase or angle of arrival, in the transmitted signal can be characterized by two main manifestations [8], large-scale and small-scale fading. Large-scale fading represents the path loss due to motion over large areas such as hills, forests, etc… [8]. The path loss can be computed as a function of distance between the transmitter and the receivers by statistical method, which is described in terms of mean-path loss and a log-normally distributed variation about the mean [8]. Small-scale fading is also called Rayleigh fading since multiple paths exists between the sender and the receiver and there’s no line-of-sight signal component, so the envelope of the received signal can be described by a Rayleigh pdf expressed as: r r2 for r ≥ 0 2 exp − 2 p ( r ) = σ 2σ otherwise 0 (1) where r is the envelope of the received signal, 2 2 is the prediction mean power of the multipath signal [8]. The Rayleigh faded component or non-specular component is also called scatter, random or diffuse component. When there is a predominant line-of-sight (specular) component, the envelope amplitude of the received signal can be statistically described by a Rician pdf [9]. Based on [8], small-scale fading manifest itself in two mechanisms: signal dispersion caused by time spreading of digital pulses within the signal, and, time variance of the channel due to the relative motion between the sender and the receiver. Furthermore, these manifestations can be categorized in various degradations types as Fig. 3 illustrates. From a time domain point of view, a channel exhibits frequency-selective fading when the maximum excess delay time is greater than the symbol time. In other words, when the maximum spread in time of a symbol is greater σ than the duration of the symbol. This is also known as channel-induced ISI since it yields to inter-symbol interference distortion [8]. From a frequency domain point of view, frequency-selective fading occurs when some spectral components of a signal, are affected in a different way than the rest of components. When the above conditions are not met, a channel is said to exhibit flat fading. Similarly, from a time domain point of view, a channel is considered to be a fast fading channel when the time duration in which the channel’s response is invariant, or coherence time is greater than the time duration of a transmission symbol. This may cause distortion of the shape of the baseband pulse. A fast fading channel, from a frequency point of view, can be described as a channel in which the symbol rate (inverse of spread time of a symbol) is less than the fading rate (inverse of the coherence time). On the other hand, a slow fading channel is a channel in which the aforementioned conditions are unmet. 3. SIMULATIONS AND RESULTS Fig. 4 shows the Simulink model used to investigate the effect of wireless channel on clipping distortion. The model is composed of several bocks. A sampled read from workspace block is used to create a data vector of size 10000, by executing the RANDINT Matlab function, which generates in our case, 10000 random integers between 0 and 3. The data output sample time was set to 10-4 sec. In other words, 10000 random integers were generated over 1 sec time duration. The data vector was used as an input for the QPSK modulator block, which outputs a baseband representation of the modulated signal. The phase offset and the samples per symbol were set to 0, 1 respectively. After the signal is modulated, it passes through the Rayleigh fading channel to simulate a wireless channel with no line-of-sight signal component. Table 1 presents the parameters that define this channel. Parameter Doppler Frequency Value 40Hz Sample Time -1 Delay Vectors [0] s Gain Vectors Initial Seed [0] dB 1234 Description The maximum Doppler shift, apparent change in frequency due to the relative motion of the sender and the receiver The period of each element of the input signal. Inherited from previous block (10-4 s) Specifies the propagation delay for the path. In this case, we assume that there’s only one path whose delay is zero Specifies the gain for the path. We assume that it’s zero The scalar seed for the Gaussian noise generator. Table 1. Simulation parameters defining the Rayleigh fading channel A bias current (we assume, the bias current has a zero intensity throughout the computer simulations) is then added to faded signal, and the magnitude of the resulting signal is plotted versus the time as shown in Fig. 5. We assume that our threshold value is 2.0V, therefore the signal will be clipped if it goes above 2.0V. Fig. 6. shows a histogram of the number of samples for different voltage levels. It can be concluded that for this fading channel, 159 samples out of 10000, or 1.59% of samples will be cut off or clipped. 2.5 3.5 2.0V 2.1V 2.2V 2.3V 2.4V 2.5V 3 threshold 2 Clipping Percentage Signal Magnitude (V) 2.5 1.5 1 2 1.5 1 0.5 0.5 0 0 0.1 0.2 0.3 0.4 0.5 Time (s) 0.6 0.7 0.8 0.9 1 Fig. 5. Magnitude of faded signal as a function of time (1 path with 2.0V as threshold voltage) 1000 1 2 3 4 Number of Fading Paths in the Channel 5 Fig. 7. Clipping percentage versus the number of paths in the fading channel for various current thresholds. Furthermore, we observed that by increasing the threshold voltages by 0.5V in laser diode, the clipping percentage may be reduced by as much as 1.75%. However, this has a drawback since an increase in the threshold voltage, can reduce the stimulated emission region’s range, which will eventually affect the efficiency of current-to-optical power conversion in the laser diode. 900 800 700 Number of Samples 0 600 500 10. REFERENCES 400 300 [1] Keiser, G., Optical Fiber Communications, McGraw-Hill Higher Education, 3rd Edition, USA, 2000. 200 100 0 −0.5 0 0.5 1 1.5 2 2.5 Signal Magnitude (V) Fig. 6. Histogram of samples for different voltage levels (1 path with 2.0V as threshold voltage) Similar experiments were conducted for fading channels with 2, 3, 4, 5 paths, whose time delays are 1x103 , 2x10-3, 3x10-3, 4x10-3 s and gains are -1, -2, -3, -4 dBs respectively. The above steps were repeated for various threshold voltages. The results are shown in Fig. 7. 4. DISCUSSIONS AND SUMMARY Based on the results shown in Fig. 7., we can conclude that, for channels with four or less fading paths, a major clipping percentage variation is observed, whereas a minor variation is noted for channels with 4 or more paths. [2] Forouzan, B.A., Data Communications and Networking, McGraw-Hill Higher Education, 2nd Edition, New York, USA, 2001. [3] S Lai, and J Conradi, “Theoretical and Experimental Analysis of Clipping-Induced Noise in AM-VSB Subcarrier Multiplexed Lightwave Systems,” Journal of Lightwave Technology, vol. 15, no. 1, pp. 20-30, Jan. 1997. [4] A. R. S. Bahai, M. Singh, A. J. Goldsmith, and B. R. Saltzberg, “A New Approach for Evaluating Clipping Distortion in Multicarrier Systems,” IEEE Journal on Selected Areas in Communications, vol. 20, no. 5, pp. 1037-1046, Jun. 2002. [5] A. J. Rainal, “Limiting Distortion of CATV Lasers,” Journal of Lightwave Technology, vol. 14, no. 3, pp. 474-479, Mar. 1996. [6] K., -P. Ho, and J. M. Kahn, “Optimal Predistortion of Gaussian Inputs for Clipping Channels,” IEEE Transactions on Communications, vol. 44, no. 11, pp. 1505-1513, Nov. 1996. [7] L. E. West, “On Determining the Optimum Modulation Index for Reverse Path Lasers in Hybrid Fiber/Coax Networks,” IEEE Photonics Technology Letters, vol. 8, no. 11, pp. 15551557, Nov. 1996. [8] B. Sklar, “Rayleigh Fading Channels in Mobile Digital Communication Systems, Part I: Characterization,” IEEE Communications Magazine, pp. 90-100, July 1997. [9] Rappaport, T. S., Wireless Communications, Prentice Hall, Upper Saddle River, NJ, USA, 1996.