ECE 6640 Digital Communications Dr. Bradley J. Bazuin Assistant Professor Department of Electrical and Computer Engineering College of Engineering and Applied Sciences Chapter 5 Chapter 5: 5.1 5.2 5.3 5.4 5.5 5.6 ECE 6640 Carrier and Symbol Synchronization Signal Parameter Estimation Carrier Phase Estimation Symbol Timing Estimation Joint Estimation of Carrier Phase and Symbol Timing Performance Characteristics of ML Estimators Bibliographical Notes and References Problems Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 290 290 295 315 321 323 326 327 2 Synchronization Issues • There are times when the coherence assumption is invalid. – The channel can introduces random changes to the signal as either a random attenuation or a random phase shift. Chapter 10 deals with equalization and Chapter 13 deals with fading channels. – Alternately, imperfect knowledge of the signals at the receiver arises when the transmitter and the receiver are not perfectly synchronized. It can use only signals in the form of {sm(t − td )}, where td represents the time difference and is model as a random variable or as a random received signal phase. ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 3 Signal Parameter Estimation • Due to the unknown “time-of-flight” of the RF signal. • Due to the potential system digital sampling times compared to the “optimal” sampling times. – a non-integer number of clocks or samples per symbol can easily exist. – system time is based on “internal” clock oscillators that are never “perfect”. • We must be able to estimate and correct for carrier frequency offsets and symbol timing offsets. • A means to estimate these parameters is required for optimal processing … coherent detection. ECE 6640 4 Phase Parameter • The received signal is based on time, time offset and phase offset. – To simplify the notation, we let θ denote the parameter vector {φ, τ}, so that s(t; φ, τ) is simply denoted by s(t; θ). • Two criteria that are widely applied to signal parameter estimation: the maximum-likelihood (ML) criterion and the maximum a posteriori probability (MAP) criterion. – In the MAP criterion, the signal parameter vector θ is modeled as random and characterized by an a priori probability density unction p(θ). – In the maximum-likelihood criterion, the signal parameter vector θ is treated as deterministic but unknown. ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 5 Phase Estimation • The joint PDF of the random variables (r1 r2 · · · rN ) in the expansion can be expressed as p(r|θ). – Then, the ML estimate of θ is the value that maximizes p(r|θ). – The MAP estimate is the value of θ that maximizes the a posteriori probability density function – if there is no prior knowledge of the parameter vector θ, we may assume that p(θ) is uniform (constant) over the range of values of the parameters. In such a case, the MAP and ML estimates are identical. – In the textbook treatment of parameter estimation given below, we view the parameters φ and τ as unknown, but deterministic. Hence, we adopt the ML criterion for estimating them. ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 6 ML Estimation • In the ML estimation of signal parameters, we require that the receiver extract the estimate by observing the received signal over a time interval T0 ≥ T , which is called the observation interval. – One shot estimation may be useful and sufficient. – Typically, estimation is performed continuously. • It is possible to derive the parameter estimates based on the joint PDF of the R.V. but it is convenient to deal directly with the signal waveforms – develop a continuous-time equivalent ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 7 ML Estimation • The pdf is – where – expressed in terms of the time signals – maximization of p(r|θ) with respect to the signal parameters θ is equivalent to the maximization of the likelihood function. ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 8 Example of Receivers with Parameter Estimation • • • • BPSK: Figure 5.1-1 MPSK: Figure 5.1-2 M-ary PAM: Figure 5.1-3 QAM: Figure 5.1-4 • Notice the similarity of functional diagrams. • Significant processing is implied in the receivers in addition to symbol filtering/correlation and detection ECE 6640 9 BPSK and MASK Receiver ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 10 MPSK and QAM Receiver ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 11 Carrier Phase Estimation Two approaches: • Embedded a carrier in the waveform for detection and estimation – time multiplexed with data symbols – a pre-amble or post-amble to a burst – use a phase-locked loop (PLL) to lock onto the carrier • Derive the carrier phase estimate from the symbols – does not require multiplexing or non-data signaling – more complex, but usually the performed method ECE 6640 12 Carrier Phase Errors • QAM and PSK Signal representation – Mixing by a carrier with phase error – Results in in-phase and quadrature symbols with phase offsets • The offset misadjust detected signal vectors. There is a reduction in the desired signal component value and a potential for cross-talk between the in-phase and quadrature-phase components. ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 13 ML Carrier Phase Estimation • Assume that the delay τ is known and, in particular, we set τ = 0. The function to be maximized is the likelihood function given in Equation 5.1–8.With φ substituted for θ, this function becomes • The first term of the exponential factor does not involve the signal parameter φ. The third term, which contains the integral of s2(t; φ), is a constant equal to the signal energy over the observation interval T0 for any value of φ. Only the second term, which involves the cross correlation of the received signal r (t) with the signal s(t; φ), depends on the choice of φ. • ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 14 ML Carrier Phase Estimation (cont) • Therefore, the likelihood function (φ) may be expressed as two “constants” and a term of interest – Note that the “relative gain scaling” is represented as a constant as well. • As the exponential function is monotonic, we can concentrate on maximizing the argument • Moving forward an example helps to define a process … ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 15 ML Carrier Phase Est. Example • Example 5.2-1: Carrier phase estimation for an unmodulated carrier. s t , LO cos2 f c t LO r t A cos2 f c t err nt 2 r t cos2 f c t LO N 0 TC 2A L A cos2 f c t err nt cos2 f c t LO dt N 0 TC L L 2A 2A cos2 f c t err cos2 f c t LO dt nt cos2 f c t LO dt N 0 TC N 0 TC L L A 2A coserr LO cos4 f c t err LO dt n2 t dt N 0 TC N 0 TC A A 2A coserr LO dt cos4 f c t err LO dt n2 t dt N 0 TC N 0 TC N 0 TC L ˆML ECE 6640 A coserr LO dt N 0 TC A N sin err LO dt 0 TC tan 1 sin err LO err LO tan 1 A coserr LO N coserr LO dt 0 TC Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 16 Estimators: PLL or One Shot Direct Computations • Based on the error oscillator model, the following estimators can be used. ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 17 Phase Lock Loop (PLL) Operation • • PLLs are discussed in the time domain and then, with assumptions that the loop is in or near lock, analyzed in the Laplace domain. Time domain first … ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 18 PLL Laplace Domain • When the loop is locked, it can be analyzed in the phase or phase error domain as a linear system. • In integrator in the feedback loop allows errors to be zeroed in the steady state response of the circuit. ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 19 PLL Loop Filter Based Responses ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 20 Noise in PLL • The PLL approach functions in a sufficiently high SNR environment. In this region, there is a derived noise term to the estimate that is presented and discussed. ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 21 Decision-Directed Loops • When the signal carries the information sequence, the estimation process is more complex. – We must either use the probabilistic properties of the information. or – We have to assume symbols are known or correctly detected (thereby the reference to decision directed). • If we know the information, ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 22 Decision-Directed Loops • This approach has isolated the phase term so that • It is easily shown (Problem 5.10) that the mean value of φML is φ, so that the estimate is unbiased. Furthermore, the PDF of φML can be obtained (Problem 5.11) by using the procedure described in Section 4.3–2. ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 23 PAM Decision-Directed Loop ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 24 QAM Decision-Directed Loop ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 25 MPSK Decision Directed PLL This M-phase tracking loop has a phase ambiguity of 360/M, necessitating the need to differentially encode the information sequence prior to transmission and differentially decode the received sequence after demodulation to recover the information. ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 26 Matlab Simulation • MATLAB comm.CarrierSynchronizer System object • The CarrierSynchronizer System object™ compensates for carrier frequency and phase offsets for single carrier modulation schemes. – [1] Rice, Michael. Digital Communications: A Discrete-Time Approach. Upper Saddle River, NJ: Prentice Hall, 2009, pp. 359–393. – [2] Huang, Zhijie, Zhiqiang Yi, Ming Zhang, and Kuang Wang. "8PSK Demodulation for New Generation DVB-S2." International Conference on Communications, Circuits and Systems, 2004. ICCCAS 2004. Vol. 2, 2004, pp. 1447–1450. ECE 6640 MATLAB Material, see http://www.mathworks.com/help/comm/ref/comm.carriersynchronizer-class.html 27 Non-Decision Directed Loops • • The first approach described assumes symbol synchronization … The squaring loop (or nth power loop) functions by taking the input to the nth power where any symbol phase ambiguity would be removed. An n x fc frequency reference is formed and divided by n before being used for coherent signal demodulations. ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 28 Costas Loop • Another method for generating a properly phased carrier for a doubles ideband suppressed carrier signal is illustrated by the block diagram shown in Figure 5.2–15. This scheme was developed by Costas (1956) and is called the Costas loop. – Highly used and useful in analog communications for carrier phase recovery. This “structure” has already been seen in the other approaches! ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 29 Symbol Timing Estimation • In a digital communication system, the output of the demodulator must be sampled periodically at the symbol rate, at the precise sampling time instants tm = mT+τ , where T is the symbol interval and τ is a nominal time delay that accounts for the propagation time of the signal from the transmitter to the receiver. • To perform this periodic sampling, we require a clock signal at the receiver. The process of extracting such a clock signal at the receiver is usually called symbol synchronization or timing recovery. ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 30 Maximum-Likelihood Timing Estimation For PAM, QAM and MPSK ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 31 Early-Late Gate Synchronization • Another non-decision-directed timing estimator exploits the symmetry properties of the signal at the output of the matched filter or correlator. – The output of the filter matched to s(t) attains its maximum value at time t = T. – Sample early, at t = T−δ and late at t = T + δ. The absolute values of the early samples |y[m(T − δ)]| and the late samples |y[m(T + δ)]| will be smaller than the peak. Use them to align the time delay. ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 32 Early-Late Gate Diagram • The early–late gate synchronizer described above is a nondecision-directed estimator of symbol timing that approximates the maximum-likelihood estimator. ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 33 Joint Estimation of Carrier Phase and Symbol Timing • Based on the similarity of structures the processes can (should) be merged. ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 34 Performance ECE 6640 Notes and figures are based on or taken from materials in the course textbook: J.G. Proakis and M.Salehi, Digital Communications, 5th ed., McGraw-Hill, 2008. 35