OVERVIEW OF MULTI-CARRIER AND MULTI-CODED SPREAD SPECTRUM SYSTEMS

Transcription

1 Chapter - 2 OVERVIEW OF MULTI-CARRIER AND MULTI-CODED SPREAD SPECTRUM SYSTEMS This chapter aims to give an overview of the principles of multi-carrier and multi-coded spread spectrum systems, which are made use of in this study. These systems are used in software simulation and so some aspects of satellite communication system simulation are also mentioned here. 2.1 Introduction Spread spectrum communications grew out of research efforts during World War II to provide secure means of communication in hostile and military environments. This work remained classified, for the most part, until the 1970s. It can be said that the technology became accessible to public with the first special issue of the IEEE Transactions in communications on spread spectrum in This was followed by numerous papers on every aspect of the spread spectrum technology, with the popular one of Viterbi [1979, reprinted again in 2002] and Marvin et al [1985], which has helped in the popularity and development of new applications for spread spectrum techniques. The type of spread spectrum used in this work is the most popular one also, the Direct Sequence Spread Spectrum (DSSS) systems. The DSSS process spreads the data message to a much wider bandwidth using spreading signals called as pseudorandom codes, therefore reducing its power spectral density. The spreading signal appears similar to Gaussian noise and hence it has a low probability of intercept (LPI) and anti-jam (AJ) features, as detailed by Dixon [1976], Skaug and Hjelmstad [1985], Zeimer et al [1985] and Proakis [2001]. For a doubly dispersive channel as assumed for satellite communication, a BPSK modulated DSSS signal can be mathematically represented as 2 (2.1) where P is the transmitted power and ω c is the carrier frequency. The spreading code is given by a 1 (t) and the data by b(t). 8

2 The received signal after propagating over a fading channel as given by Gardner and Orr [1979] is 2 cos (2.2) where f (t- ) represents the fading process introduced by the channel with τ as the propagation delay, θ is the carrier phase and n(t) is additive white noise. Detection depends on the realization of the matched filter. This is normally accomplished by multiplying a band limited version of the received signal with a local oscillator modulated by the time spreading code as reported by Pijoan et al [1999]. For digital implementations it is usually required to low pass filter the received signal, and then use in phase and quadrature techniques on a sampled and quantized version of the input signal. Here, the nonlinearities involved in the quantization are neglected and satisfactory sampling is assumed. The output for a single data bit duration T d, then becomes cos (2.3) The second term is the noise with n (t) assumed as a zero-mean Gaussian process. 2.2 A review of some properties of PN codes The pseudo-random (PN) codes, which form the basis of spreading the spectrum in a DSSS system has several important characteristics, which are to be followed to develop a spread spectrum system satisfying the requirements. The PN codes are designed according to the applications and much literature is available for this in Gold [1967], Ziemer and Peterson [1985], Dixon [1976], Sarwate and Pursley [1980], Jack Holmes [1982], Sarwate et al [1984], Skaug et al [1985], Mahalakshmi and Karunakaran [2003], to name a few. The main criterion for selecting a set of codes is based upon the autocorrelation and cross-correlation functions of these codes. The autocorrelation function, defined as the scalar product of the code signal with code signal shifted by a delay τ, should be ideally zero for τ 0. Such a good autocorrelation property is useful in satellite and 9

3 mobile communication systems for separating the different propagation paths and hence avoiding inter symbol interference, leading to proper timing recovery and coherent detection as explained by Jack Holmes and Chang Chen [1977] and Jack Holmes [1982]. In spread spectrum multiple access systems which use multiple codes as in Code Division Multiple Access (CDMA) systems, different code signals C and C are used to distinguish different channels. The mutual interference between these channels is proportional to the scalar product of C and C as detailed by Garber and Pursley [1980], Kaiser [1995, 2002], Popovic [1999] and Fazel and Kaiser [2008]. Hence, for these applications, one important property is to have orthogonality of codes, which means the cross-correlation function of C and C has to be minimized. Each of the known types of codes fulfills one requirement to a higher and the other to a lower degree. Therefore, the codes giving the best combination for the respective application has to be selected. The codes mentioned below are either used for the simulation studies in this work or are presently used for satellite applications Maximal Length Sequence (MLS) codes or m sequences These codes have normally a good autocorrelation property and can be generated by linear feedback shift registers. A linear feedback shift register of length m produces an m- sequence if and only if the corresponding generating polynomial of degree m is primitive. A polynomial of degree m is called primitive if it is irreducible, that is, if it cannot be factorised. e.g. P(X) =X 5 + X Hence, a code given by an m- sequence has nearly an ideal autocorrelation function for large M values. However, the cross correlation peak values decreases quite slowly with increasing sequence length M. As this type of code is easy to generate, they are used for the simulation studies attempted as part of this work Gold codes A set of Gold codes of length M can be obtained by combining specific pairs of m- sequences like C, C which are called preferred m- sequences. For large M the peak values of the cross- correlation functions of Gold codes are much smaller than for 10

4 the m- sequences, but at the expense of higher and decreasing values of the autocorrelation functions. The combined codes in the set of Gold codes are no m- sequences and have favourable cross-correlation properties. These codes are used in GPS systems, the details of which are studied as part of this work. 2.3 Features of Multi carrier and Multi-coded SS signals Any spread spectrum system has three major signal components data, PN code and carrier. In order to multiplex different spread spectrum signals to optimize spectrum usage, it is possible to use either multiple carriers or multiple codes or multiple code-carrier combinations, depending on the application. In this work, use of all the three for TEC measurement has been examined through simulation and/or measurement. The following section looks into the basic concepts behind each of these techniques. When there is a need to transmit more than one DSSS over a single channel, the two popular variants are Code Division Multiple Access (CDMA) and Orthogonal Frequency Division Multiplexing (OFDM) type. Code division multiplexing (CDM) allows signals from a series of independent sources to be transmitted at the same time over the same frequency band, as detailed by Tiedemann et al [1991], Gilhousen et al [1991]. This is accomplished by using orthogonal codes to spread each signal over a large, common frequency band. Thus it can be considered as a special case of multi-coded spread spectrum system. At the receiver, the appropriate orthogonal code is then used to recover the particular signal intended for a particular user. The orthogonal nature of the codes ensures that multiple users communicate over the channel with minimal interference. The popular orthogonal codes with well defined constructions include Walsh-Hadamard codes and Orthogonal Gold codes. The number of codes in an orthogonal set is limited to the spreading factor. Channel estimation, equalization and power-control procedures are necessary in a DS-CDMA system, which makes the receiver more complex to design. In present day GPS systems and possibly in future beacon satellite systems, when multiple satellites are simultaneously visible above a single ground receiver station, 11

5 each satellite can be uniquely identified by its specific code. The following figure 2.1 explains the encoded format of a sample message by different PRN codes. Figure 2.1 Example of CDMA coding technique Multicarrier modulation, in the most general sense, can refer to any modulation scheme that uses multiple carrier frequencies to transmit data as mentioned by John Bingham [1990] and Reiners and Rohling [1994]. It can be considered as a spread spectrum modulation technique in which each bit is modulated on multiple subcarriers with relative phase polarity according to a spreading code, according to Fazel and Kaiser [2008]. Multicarrier techniques have been developed since as early as the sixties. Towards the end of the sixties, a number of authors, notably Chang [1966], used overlapping orthogonal spectra to increase the efficiency of multicarrier systems. Orthogonal Frequency Division Multiplexing (OFDM) is a special form of multicarrier modulation, which was patented in 1970 and described later on by Yiyan Wu and Zou [1995], Kamuang et al [2000], Charan Langton [2002]. It is well suited for transmission over a dispersive channel as detailed by Jean-Paul M.G. Linnartz [1993]. Basically, in an OFDM system, the total bandwidth B is divided into K sub-bands with orthogonal subcarriers. For a total block of M symbols to be transmitted, each block has 1/M of the available bandwidth instead of just transmitting one symbol as in single-carrier data 12

6 transmission as detailed in the work of Karp [2001]. OFDM transmission techniques are widely deployed for WLANs, ADSL, VDSL modems, etc., but a standard using this technique for satellite communications still does not exist. The main difficulty in using the system is the complexity of the equipment needed for its implementation, since mutual interference between the subcarriers is possible and filters of accurate cut-off frequencies are needed. OFDM technique is able to address multipath fading as it can efficiently deal with channel delay spread and is robust to narrowband interference. But it is sensitive to small carrier frequency offsets and sampling clock offsets. MC-CDMA combines the benefits of CDMA with the natural robustness to frequency selectivity offered by OFDM as detailed by Prasad and Hara [1996]. It can be interpreted as CDMA with the spreading taking place in the frequency rather than temporal domain. In MC-CDMA, the processing and signature spreading occurs in the frequency domain. A comparison of the spectra of BPSK, CDMA and MC-CDMA signals is shown in figure 2.2 where it can be seen that the signal spread is rather large for DS-CDMA systems while for MC-CDMA systems, each of the subcarrier has a spectrum similar to DSSS-BPSK. Figure 2.2 Comparison of spectra A typical time-amplitude spectrum of OFDM and MC-CDMA is shown in figure 2.3. Here, it can be seen that MC-CDMA has both positive and negative values 13

7 whereas OFDM signal has only positive values, with good overlap between adjacent sub-carriers. Figure 2.3 Time-amplitude spectra of OFDM and MC-CDMA 2.4 A brief introduction to simulation of communication systems As mentioned above, simulation of spread spectrum beacon systems have been attempted as part of this study. In this context, a brief introduction on communication system simulation is given here. A large body of computer- aided techniques has been developed in recent years to assist the process of modeling, analyzing, and designing communication systems. These computer aided techniques fall into two categories as indicated by Schiff [2006]: formula-based approaches, where the computer is used to evaluate complex formulas, and simulation-based approaches, where the computer is used to simulate the waveforms or signals that flow through the system. Formula based techniques, which are based on simplified models, provide considerable insight into the relationship between design parameters and system performance, and they are useful in the early stages of the design for broadly exploring the design space. With simulation based approaches of performance evaluation, systems can be modeled with almost any level of detail desired and the design space can be explored more finely. In this latter approach, one can combine mathematical and empirical models easily, and incorporate measured characteristics of devices and actual signals into 14

8 analysis and design. Simulated waveforms can also be used as test signals for verifying the functionality of hardware. Thus simulation can be defined as the process of designing a model of a real system and conducting experiments with this model for the purpose of either understanding the behaviour of the system and/or evaluating various strategies for the operation of the system. As told by Tranter and Kosbar [1994], simulation processes start with formulating the problem and planning the study. This is followed by collecting relevant data to generate and define a model. The next step is to identify a computer language which can be used to develop the simulation software suited to the application. For this thesis work, the LabVIEW programming language is chosen for simulation software design. LabVIEW (short for Laboratory Virtual Instrumentation Engineering Workbench) is a platform independent development environment for a visual programming language from M/s. National Instruments. The programming language used in LabVIEW, referred to as G, is a dataflow programming language. Originally released for the Apple Macintosh in 1986, LabVIEW is commonly used for data acquisition, instrument control, and industrial automation on a variety of platforms including Microsoft Windows, Linux, and Mac OS X. The fully object-oriented character of LabVIEW code allows code reuse without modifications i.e., as long as the data types of input and output are consistent, any two subvis are exchangeable. Many libraries with a large number of functions for data acquisition, signal generation, mathematics, statistics, signal conditioning, analysis, etc., along with numerous graphical interface elements are provided in several LabVIEW package options. It also offers extensive support for accessing instrumentation hardware. LabVIEW programs/subroutines are called virtual instruments (VIs) with the code files having the extension.vi. Each VI has three components: a block diagram, a front panel, and a connector panel. The last is used to represent the VI in the block diagrams of other calling VIs. Controls and indicators on the front panel allow the user to input data into or extract data from a running virtual instrument. The front panel also serves as a programmatic interface. This implies each VI can be easily 15

9 tested before being embedded as a subroutine into a larger program. The program execution is determined by the structure of a graphical block diagram (the LVsource code) on which the programmer connects different function-nodes by drawing wires. These wires propagate variables and any node can execute as soon as all its input data become available. Since this might be the case for multiple nodes simultaneously, G is inherently capable of parallel execution Various sub-systems in spread spectrum system simulation The standard description in simulation of a system is a block diagram where each block represents a signal processing operation. According to Sharif and Gholampour [2011], the block diagram is a signal flow diagram indicating the generic type of operations that the signal and/or noise that drive the system are subjected to. The following are the major blocks used in the simulation for this work Information sources The stimuli or driving functions in communication systems are the outputs of various sources of information, noise and interference. Outputs of these sources may be random processes or deterministic functions and they may be analog or digital in nature. Noise and interference represent the undesirable components of a waveform. Noise arises due to natural causes and interference is manmade. In the present work of simulation, the system assumes that all frequency sources are ideal and so noise is added separately. In the simulation platform LabVIEW, a range of Virtual Instruments (VIs) are available for generation of various types of signal and noise. The control parameters to generate these signals are given according to requirement Filters There is often the need in simulation to use certain idealized forms of filtering as an approximation to reality. Filters are designed by varying their specifications like amplitude, gain slope, amplitude and phase ripple, filter type, order etc. In order to avoid aliasing in digital systems, it is understood that these filters are to be used in conjunction with a band-limiting filter that defines a bandwidth appropriate for the simulation sampling rate. 16

10 Communication channels and Models In its most general sense the word channel can be used to mean everything between the source and the sink of a signal. The channel accounts for propagation effects such as ordinary 1/R 2 free space loss, rain absorption, multipath, diffraction, refraction, and scattering, as well as general background noise. In order to simulate such a channel when it is explicitly represented by a sequence of blocks we need to have a model for each block as well as for the medium. The type of channel model, appropriate to digital transmission is called a discrete channel. In the present work, the intervening discrete channel is considered to be of Ionospheric phase channel type, which also assumes there is no multipath or scintillation. Assuming the absence of anomalous conditions (solar flares, nuclear events, extremely low elevation angles), for VHF and higher frequencies, the ionosphere can be approximately modeled by an all-pass filter with a non-ideal phase characteristics, normally considered as the ionospheric phase channel. Accordingly, a parabolic phase (Linear Time Delay) filter given by exp (2.4) where θ p is the parabolic phase (radians) at a reference frequency of ω = 1 rad/sec is used in both the simulation techniques for this work. It can be shown that the phase shift experienced by a wave of frequency f due to free electrons in the ionosphere, over and above the free- space propagation lag, is given by (2.5) where c is the speed of light (cm/sec), N e is the electron concentration per unit area (electrons/cm 2 ) at any point along the path s, and the integral represents the integrated columnar electron density along the signal path, otherwise known as Total Electron Content. The differential phase shift between any two frequencies f 0 and f 0 + f is therefore given by 17

11 Δ Δ 2π Δf cf f f, N s ds (2.6) In a multipath environment, the received signal is composed of several time delayed versions of the transmitted signal. These signals add at the receiver with different phases as a result of different propagation delays. It has been shown by Jak [1974] that the envelope of this follows a Rayleigh distribution. Additionally the rate at which the signal varies is inversely proportional to the Doppler spread of the channel. In this model we use a Rayleigh function and multiply it with the transmitted signal and then compensate for this distortion in the receiver Receiver systems In coherent communication systems, the demodulator (receiver) needs a local carrier reference whose phase is a close approximation to the phase of the incoming carrier. Furthermore, if the information transmitted is digital in nature then the receiver also needs to generate a symbol timing (clock) signal to control the sampling times at which the matched filter output is to be sampled. The process of generating carrier and timing references at the receiver is referred to as synchronization and is abided by in the simulation study in this work. 2.5 Summary This chapter summarizes the different features of multi-carrier and multi-coded spread spectrum systems relevant to this thesis work. These techniques are used as part of software simulation in this study. Some advantages of the software platform used (LabVIEW) is mentioned. Details on major sub-systems of a satellite communication system for simulation is also addressed along with details on the channel model chosen for this study. 18