A SEMINAR REPORT Submitted by SAM P JOSEPH In partial fulfilment for the award of the degree Of Master of Technology IN SIGNAL PROCESSING
DEPARTMENT OF ELECRONICS AND COMMUNICATION GOVERNMENT ENGINEERING COLLEGE, BARTON HILL KERALA UNIVERSITY, THIRUVANANTHAPURAM SEPTEMBER, 2014 GOVERNMENT ENGINEERING COLLEGE, BARTON HILL DEPARTMENT OF ELECTRONICS AND COMMUNICATION ENGINEERING BONAFIDE CERTIFICATE Certified that this seminar report AN ANALOG JOINT SOURCE CHANNEL CODING (JSCC) SYSTEM FOR WIRELESS OPTICAL COMMUNICATIONS is the bonafide work of SAM P JOSEPH, submitted in partial fulfilment of the requirements for the award of Master of Technology Degree in SIGNAL PROCESSING of the Kerala University, Thiruvananthapuram. Dr. N Vijayakumar HEAD OF DEPARTMENT Dept of ECE Government Engineering College Barton Hill Sindhu N SEMINAR COORDINATOR Assistant Professor Dept of ECE Government Engineering College Barton Hill ABSTRACT Free Space Optical communication links transmit information by laser light through an atmospheric channel. They operate at frequency bands (around 300 THz) where the spectrum is unlicensed. There is always a growing demand for optical communication both in analog and digital system. Also an effective coding technique for optical communication is less well developed. An Analog J oint Source Channel Coding (J SCC) system is developed for wireless optical communication. Source symbols are mapped directly onto channel symbols using space filling curves and then a non-linear stretching function is used to reduce distortion. Different from digital systems, the proposed scheme does not require long block lengths to achieve good performance reducing the complexity of the decoder significantly. The paper focuses on intensity- modulated direct-detection (IM/DD) optical wireless systems, where information is modulated on the intensity of light wave and direct detection measures the light wave power, which is similar to a square law detector in RF communications. The nonlinearities of the real channel are studied and characterized. The objective in this paper is to investigate the application of the Shannon Mappings in wireless optical communications. A new technique that involves a nonlinear transformation is introduced to the system model to improve its performance. The proposed system presents an excellent performance, while requiring very small complexity. iii ACKNOWLEGDEMENT First and foremost I thank GOD ALMIGHTY for giving me the strength and confidence to successfully complete this endeavor on time. I am extremely grateful to a lot of people for the successful completion of this seminar and I feel that it is my responsibility to thank those who have put in so much time and effort to make my endeavor successful. Hence I would like to acknowledge their sincere support and co-operation. I am grateful to Dr. B. Anil, Principal, Govt. Engineering College, Barton Hill for providing me the best facilities and atmosphere for the development and presentation of my seminar. I thank Prof. Dr N Vijayakumar, Head of the Department, Department of Electronics Engineering, who has been a perpetual source of encouragement and support. I am indebted to Prof. Sindhu N, Assistant Professor in Electronics and Communication Engineering, for her valuable suggestion and ideas without which this report would have been a tough task. A sincere word of thanks to all my friends and family members for their support and prayers offered which were inevitable for the successful completion of the seminar. SAM P J OSEPH iv TABLE OF CONTENTS Abstract iii Acknowledgement iv List of Figures vi 1. Introduction 1 2. Free space optics communication system 2 2.1 Free space optics (FSO): Challenges 3 3. Optimal mappings for joint source channel coding 5 4. Analog Joint Source Channel Coding (JSCC) 6 4.1 Analog J SCC using Shannon mappings 7 5. System model for AWGN channels 8 6. Channel analysis and characterisation 10 6.1 Channel attenuation and non linearities 10 6.2 Channel noise 11 7. Conclusion 12 References 13 v LIST OF FIGURES Fig.4.1 Archimedes Spiral 7 Fig.5.1 System model for AWGN channels 8 Fig.5.2 Prototype to transmit data using analog J SCC 9 Fig. 6.1 Electrical equivalent of the system 10 Fig. 6.2. (a) Channel attenuation VADC / VDAC 11 Fig. 6.2. (b) Noise pdf for VDAC =0.015 V 11 Fig. 6.2. (c) Noise pdf for VDAC =0.75 11 Fig. 6.2. (d) Noise pdf for VDAC =1.50 V. 11 vi An Analog J oint Source Channel Coding (J SCC) System For Wireless Optical Communications CHAPTER 1 INTRODUCTION Our current "age of technology" is the result of many brilliant inventions and discoveries, but it is our ability to transmit information, and the media we use to do it, that is perhaps most responsible for its evolution. Despite new technologies, optical network remains the most ideal medium for high bandwidth communications for true connectivity. Fiber Optic connections are considered very secure. Data is transmitted as beams of light so no electromagnetic waves are generated. There are two distinct types of optical communication Fiber optics Optical wireless based on FSO technology Today optical fiber carries terabits-per-second to a vast host of users who are interested in information in the form of voice, fax, video, documents, and web pages. But optical fiber represents the upper limit of available bandwidth and high data transfer rates and it also requires time to install the optical cables. Free-space optical (FSO) communication technology can provide high data rate transfer and can be easily installed, moved or reconfigured as needs change. FSO technologies are intrinsically secure because of the line-of-sight requirement as well as the high directivity of the optical beam. FSO first appeared in the 60's, for military applications. At the end of 80's, it appeared as a commercial option but technological restrictions prevented it from success. Low reach transmission, low capacity, severe alignment problems as well as vulnerability to weather interferences were the major drawbacks at that time. Free-Space Optics (FSO) is a fibreless, laser-driven technology that supports high bandwidth, with easy to install connections for the last-mile and campus environments. It has been in use by the military for a number of years primarily in naval ship-to-ship communications. Free-Space Optics systems are starting to gain acceptance in the private marketplace as a solution to replace expensive fiber optic based solutions. Dept. of Electronics and Communication 1 An Analog J oint Source Channel Coding (J SCC) System For Wireless Optical Communications CHAPTER 2 FREE SPACE OPTICS COMMUNICATION SYSTEM
Free-space optical communication (FSO) [7] systems (in space and inside the atmosphere) have developed in response to a growing need for high-speed and tap-proof communication systems. Links involving satellites, deep-space probes, ground stations, unmanned aerial vehicles (UAVs), high altitude platforms (HAPs), aircraft, and other nomadic communication partners are of practical interest. Moreover, all links can be used in both military and civilian contexts. FSO is the next frontier for net-centric connectivity, as bandwidth, spectrum and security issues favor its adoption as an adjunct to radio frequency (RF) communications. Free space optics or optical wireless is a telecommunication technology that uses light propagating in free space to transmit data between two points. It is a line of sight broadband communication technology that uses optical pulse modulated signals to wirelessly transmit data. Instead of the pulses of light being contained within a glass fiber, they are transmitted in a narrow beam through the atmosphere. Free space optics technology is wireless networking that uses light beams instead of radio waves; it's laser-based optical networking without the fiber optic It's based on connectivity between FSO-based optical wireless units, each consisting of an optical transceiver with a transmitter and a receiver to provide full-duplex (bi-directional) capability. Each optical wireless unit uses an optical source, plus a lens or telescope that transmits light through the atmosphere to another lens receiving the information. At this point, the receiving lens or telescope connects to a high-sensitivity receiver via optical fiber. A free-space optical link consists of 2 optical transceivers accurately aligned to each other with a clear line-of-sight. FSO has drawn attention in telecommunication industry, due to its cost effectiveness easy installation, quick establishment of communication link especially in the disaster management scenario, high bandwidth provisioning and wide range of applications. The range of frequencies where it operates makes FSO communication free from licensing. With FSO communication, maximum data transfer rates up to 2.5 Gbps is possible, unlike the maximum data transfer rates of 622 Mbps offered by RF Dept. of Electronics and Communication 2 An Analog J oint Source Channel Coding (J SCC) System For Wireless Optical Communications communication systems. Free-Space Optics (FSO) is a technology similar to fiber optic cable infrastructure except that no cable is involved. The light pulses are transmitted through the atmosphere in a small conical shaped beam by the means of low powered lasers or LEDs. Free-Space Optic installations require line-of-sight availability between the laser/receiver units which are called link heads. Free-Space Optics is a wireless technology it does not have the nasty habit of broadcasting to anybody and everybody. It instead transmits a very high frequency narrow beam of light to a specific destination. Free-Space optical links uses intensity modulation and direct detection (IM/DD), where information is modulated on the intensity of light wave and direct detection measures the light wave power, which is similar to a square law detector in RF communications. FSO involves the optical transmission of voice, video, and data using air as the medium of transmission. Transmission using FSO technology is relatively simple. An FSO link can be procured and installed for as little as one-tenth of the cost of laying fiber cable, and about half as much as comparable microwave/RF wireless systems. By transmitting data through the atmosphere, FSO systems dispense with the substantial costs of digging up sidewalks to install a fiber link. Unlike RF wireless technologies, FSO eliminates the need to obtain costly spectrum licenses or meet further regulatory requirements.
2.1 Free Space Optics (FSO): Challenges When light is transmitted through the air as in optical wireless systems like Free Space Optics (FSO), it must contend with a complex and not always quantifiable subject - the atmosphere. Free Space Optics (FSO) technology based wireless systems are not without challenges[7]. The fundamental limitation of free space optical communications arises from the environment through which it propagates. Although relatively unaffected by rain and snow, free space optical communication systems can be severely affected by fog and atmospheric turbulence. Fog is vapour composed of water droplets, which are only a few hundred microns in diameter but can modify light characteristics or completely hinder the passage of light through a combination of absorption, scattering, and reflection. This can lead to a decrease in the power density of the transmitted beam, decreasing the effective distance of a free space optical link. Dept. of Electronics and Communication 3 An Analog J oint Source Channel Coding (J SCC) System For Wireless Optical Communications There is a growing demand of optical communications for both analog and digital systems. Rigorous fundamental limits of optical channels are still unknown, therefore effective coding techniques for Optical Communications are less well developed. Hence it becomes important to study the optical channel and develop new coding schemes.
Dept. of Electronics and Communication 4 An Analog J oint Source Channel Coding (J SCC) System For Wireless Optical Communications CHAPTER 3 OPTIMAL MAPPINGS FOR JOINT SOURCE CHANNEL CODING
The Shannon separation principle [5] states that the optimal communication system can be achieved by cascading optimal source coding and optimal channel coding. It is well known that under some circumstances analog communications are optimal One of the fascinating results in information theory is that, direct transmission of uncoded Gaussian samples over AWGN channels is optimal. In that sense, it is said that Gaussian sources are perfectly matched to Gaussian channels. It also yields the minimum mean square error between source and reconstruction. This result demonstrates the potential of joint source channel coding. Such a simple scheme with no delay provides the performance of the optimal separate source-channel coding system. In the practical problem of transmitting a discrete time continuous alphabet source over a discrete time additive analog channel, there are two main approaches: analog communication through direct amplitude modulation, and digital communication which typically consists of quantization, error control coding and digital modulation. The main advantage of digital over analog communication is due to advanced quantization and error control schemes. However, there are two notable shortcomings: First, error control coding (and to some extent also source coding) usually incurs substantial delay to achieve good performance. The other problem is the level off effect due to underlying quantization. The performance saturates as channel signal to noise ratio (CSNR) increases above the threshold. Analog systems offer the potential to avoid these problems. Among the few practical analog coding schemes that have appeared in the literature are those based on the use of space-filling curves for bandwidth compression, originally proposed more than 50 years ago by Shannon. Spiral-like curves are explored for transmission of Gaussian sources over AWGN channels
Dept. of Electronics and Communication 5 An Analog J oint Source Channel Coding (J SCC) System For Wireless Optical Communications CHAPTER 4 ANALOG JOINT SOURCE-CHANNEL CODING (JSCC)
J SCC[1] has recently emerged as a coding scheme that works purely in the analog domain. In contrast to traditional digital systems where zeros and ones are coded and transmitted, analog J SCC takes analog input symbols and generates analog channel symbols without any processing in the digital domain. It is known that approaching optimality requires infinite block lengths, which results in long system delays and high computational complexity. Moreover, if the signal to be sent is analog, quasi-optimal digital systems would require powerful vector quantization and source coding. Channel codes are designed to protect data at a certain channel signal to noise ratio (CSNR). If the CSNR increases, the channel codes will over-protect the data, which leads to low efficiency. On the other hand, if the CSNR decreases, the channel codes will no longer offer sufficient protection, and this can lead to a breakdown in system performance. Our goal is to design low-complexity analog J SCC schemes that present low distortion in lossy compression and graceful degradation when the channel conditions change. Roughly speaking, source coding is a data compression process that aims at removing as much as possible redundancy from the source signal, while channel coding is the process of intelligent redundancy insertion which creates some kind of protection against the channel noise. Recent developed techniques, known as Shannon Mappings can meet our goal in both bandwidth compression and bandwidth expansion. This scheme uses space filling curves to fill an n-dimensional space with a one dimensional curve to create a mapping of R N R or R R N . By selecting the direction of mapping, one can achieve N: 1 bandwidth compression or 1: N bandwidth expansion. This scheme has shown surprisingly good performance when dealing with lossy compression in AWGN channels, with a resulting distortion that is very close to the theoretical limit.
Dept. of Electronics and Communication 6 An Analog J oint Source Channel Coding (J SCC) System For Wireless Optical Communications 4.1 Analog JSCC using Shannon mappings [1]
The idea of using geometric curves for bandwidth expansion was first introduced by Shannon in 1949. The name Shannon Mappings[1][5] was given by Ramstad in to honor the idea from Shannon 60 years ago. Akyol et al. studied the optimal mapping schemes for different bandwidth compression/expansion ratios and concluded that the Archimedes spiral is very close to the optimal mapping scheme for 2:1 bandwidth compression [5] when optimal decoders are used. In the literature, different types of curves have been proposed according to the source distribution. In general, the idea is to encode an N-tuple of i.i.d source symbols into K channel symbols and then transmit them through a channel with noise variance 2N . It is known that for i.i.d zero mean Gaussian sources 2:1 mappings using the Archimedes spiral can achieve quasioptimal performance. The Archimedes Spiral can be described by the following parametric form:
The two source symbols (x1, x2) are mapped onto the closest point on the Archimedes Spiral thereby generating the corresponding channel symbol represented by the angle .
________(4.1) Fig.4.1 Archimedes Spiral
Dept. of Electronics and Communication 7 An Analog J oint Source Channel Coding (J SCC) System For Wireless Optical Communications CHAPTER 5 SYSTEM MODEL FOR AWGN CHANNELS
Fig.5.1. System model for AWGN channels
The system model [1] shown in Fig. 5.1 is utilized to transmit the analog symbols through the AWGN channel. After performing the mapping described above, represented by M(), a nonlinear function T
(), often called stretching function is used to reduce
distortion. The transformation T 2 () =sin()||2 shows good performance close to OPTA. The parameter can be optimized together with to obtain better performance, especially when the CSNR is low. Finally at the end of the encoder the normalization constant K is used. At the decoder the received signal is first multiplied by the Constant K and then ML or MMSE decoding is applied to obtain an estimated value X ^ . When ML decoding is used X ^ is calculated by,
where X is the angle obtained after mapping X on the Archimedes Spiral and s is the spiral-like curve. Therefore ML decoding can be performed by first applying the inverse function T () 1 on the received symbol y and then the inverse mapping. In IM/DD systems the information of the signal is modulated on lightwave intensity at the transmitter and is then recovered by direct detection of its power at the receiver. This paper focuses on optical wireless channels where an electrical signal is transformed into a light beam by a photodiode and then received by a photo-detector (either a photodiode or phototransistor) which converts the received optical power back into an electric signal.
________(5.1) Dept. of Electronics and Communication 8 An Analog J oint Source Channel Coding (J SCC) System For Wireless Optical Communications
Fig. 5.2 Prototype to transmit data using analog J SCC
The optical transmitter is driven by an electrical signal x(t) (input signal) that produces the intensity I(t). This can be written as, I(t) =ax(t), where a is the optical gain of the emitter. Since direct detection is performed by our system, the electrical signal produced by the photo detector y(t) is proportional to the received intensity and can be written as y =bI(t). If we take into account the channel response, the mathematical expression that describes the received signal is y(t) =|abx(t) h(t) +n|, where the absolute values is due to the fact that intensity cannot be negative, the product ab is unitless and the function h(t) represents the impulse response of the channel [23].Where the absolute values is due to the fact that intensity cannot be negative, the product ab is unit less and the function h(t) represents the impulse response of the channel. The noise n is predominantly shot noise generated by the high intensity ambient light and by the Central Limit Theorem it can be modeled as signal-independent AWGN with variance 2 n. In our system model the signals are considered to be bandwidth-limited over the operation frequency range and the channel frequency response to be flat over that region. This leads to the following simplified channel model: y(t) =|hx(t) +n|, where h represents the product ab. Since the physical quantity x(t) is instantaneous optical power, this constrains all the transmitted amplitudes to be nonnegative
Dept. of Electronics and Communication 9 An Analog J oint Source Channel Coding (J SCC) System For Wireless Optical Communications CHAPTER 6 CHANNEL ANALYSIS AND CHARACTERIZATION
Real wireless channels and semiconductor devices present nonlinear behaviours that affect the performance of the overall system and represent a challenge at the decoding stage. Signal attenuation due to the distance between transmitter and receiver is also a factor to be taken into account since in some cases its effect can be severe and affect the signal reconstruction. For these reasons having some knowledge of the channel characteristics and behaviour becomes crucial in real communication systems.
6.1 Channel attenuation and nonlinearities:
The channel attenuation includes the semiconductor nonlinear effects in both transmitter and receiver.
Fig. 6.1 Electrical equivalent of the system The optical channel model can be represented by an equivalent electrical model. Thus, the characterization of the channel is performed measuring electrical signals. The channel attenuation coefficient h is obtained as h =VADC/VDAC, thereby h is represented as the relationship between the voltage output of the DAC that controls the photodiode and the voltage across the resistor R represented by VADC as shown in Fig. 4. For this experiment VADC is a known dc voltage signal set by the microcontroller at the transmitter and VADC is taken as the sample mean of a defined number of samples N taken by the ADC at the receiver. When VADC is low the behavior is highly nonlinear since the devices are not working entirely in their active region. However, as the VADC is increased the gain remains relatively stable for VADC >1 V. Dept. of Electronics and Communication 10 An Analog J oint Source Channel Coding (J SCC) System For Wireless Optical Communications
Fig. 6.2. (a) Channel attenuation VADC / VDAC , (b) Noise pdf for VDAC =0.015 V, (c) Noise pdf for VDAC =0.75, (d) Noise pdf for VDAC =1.50 V.
6.2 Channel Noise: The maximum likelihood estimators were used to calculate the sample mean and the variance 2 of the received signal VADC. Despite the nonlinear behaviour of the channel the noise variance remains almost constant throughout the entire dynamic range of the device. The probability density function (pdf) of the noise distributions for different values of VADC are shown in Fig. 5(b)(d). The pdf of the noise distributions for different values of VADC are shown in Fig. 5(b)(d). It can be seen that at the receiver the real noise resembles Gaussian noise, and for values close to 0 V, as in Fig. 5(b), the distribution folds as a result of the optical channel constraints. This can be modeled using a Chi distribution of order 1 (the distribution of the absolute value of a Gaussian distribution). By approximating the system noise as zero mean Gaussian noise of the same variance and taking the absolute value of the received signal, the system can be modeled accurately and without any noticeable performance degradation. Dept. of Electronics and Communication 11 An Analog J oint Source Channel Coding (J SCC) System For Wireless Optical Communications CHAPTER 6 CONCLUSION
An analog J SCC communications system was designed for the IM/DD wireless optical channel scheme for AWGN channels. Based on experimental results, obtained using our prototype communication system, the IM/DD channel is characterized and its capacity calculated. One of the goals was to study the applicability of analog J SCC in real communication systems. Taking into account the nonlinear behaviour of the channel, a novel analog J SCC scheme is proposed to transmit data over the optical channel. The performance of the proposed scheme, in theory and practice, closely approaches the theoretical limits for ML and MMSE decoding when the parameters of the system are optimized. A new technique that involves a nonlinear transformation was introduced to the system model to improve its performance. The proposed system presents an excellent performance, while requiring very small complexity. The proposed system can be used for image transmission by first taking samples of a set of images using compressive sensing and then encoding the measurements using analog J SCC.
Dept. of Electronics and Communication 12 An Analog J oint Source Channel Coding (J SCC) System For Wireless Optical Communications
REFERENCES
[1] Sergio Matiz Romero, Mohamed Hassanin, J avier Garcia-Frias, Analog J oint Source Channel Coding for Wireless Optical Communications and Image Transmission, J ournal of lightwave technology, vol. 32, no. 9, May 2014. [2] M. Gastpar, B. Rimoldi, and M. Vetterli, To code, or not to code: Lossy source- channel communication revisited, IEEE Trans. Inf. Theory, vol. 49, no. 5, pp. 1147 1158, May 2003. [3] Y.Hu, J . Garcia-Frias, and M. Lamarca, Analog joint source-channel coding using non-linear curves and mmse decoding, IEEE Trans. Commun., vol. 59, no. 11, pp. 30163026, Nov. 2011. [4] O. Freznedo, F. Vazquez-Araujo,M. Gonzales-Lopez, and J . Garcia-Frias, Comparison between analog joint source-channel coded and digital bicm systems, in Proc. IEEE Int. Conf. Commun., 2011, pp. 15. [5] F. Hekland, G. Oien, and T. Ramstad, Using 2:1 Shannon mapping for joint source- channel coding, in Proc. Data Compress. Conf., Mar. 2005. pp. 16 [6] Y. Hu and J . Garcia-Frias, Optimizing power allocation in analog joint source- channel coding, in Proc. IEEE 43rd Annu. Conf. Inf. Sci. Syst., 2009, pp. 7276. [7] H.Henniger, O. Wilfert, An introduction to FSO communication, Radio Engineering 2010.