In this paper, we offer a solution to the stochastic realization problem for a Gaussian Markov field defined on a tridimensional lattice, which is a graph with nodes regularly positioned to form a discrete parallelepiped in the euclidean space and arcs connecting “internal' nodes with five nearest neighbors along the three coordinate directions. Next we show how the stochastic realization can be used for weather forecasting via a Kalman predictor, relying on partial observations and just a purely statistic a-priori knowledge of the Markov field, similarly to a classic Hidden Markov Model (HMM). An application carried out on real climate data shows the effectiveness of the approach taken.

Future wireless networks (e.g., 5G) will consist of multiple radio access technologies (RATs). In these networks, deciding which RAT users should connect to is not a trivial problem. Current fully distributed algorithms although guaranteeing convergence to equilibrium states, are often slow, require high exploration times and may converge to undesirable equilibria. To overcome these limitations, this paper develops a network feedback framework that uses limited network-assisted information to improve efficiency of distributed algorithms for RAT selection problem. We prove theoretically that a fully distributed algorithm developed within this framework is guaranteed to converge to a set of correlated equilibria. Our framework guarantees convergence in self-play even when only a single user applies the algorithm. Simulation results demonstrate that our solution: 1) is highly efficient with fast convergence time and low signaling overheads while achieving competitive, if not better, performance both in fairness and utility, as well as achieving lower per-user switchings than state-of-the-art algorithms; and 2) can flexibly support a wide range of network-assisted feedback. The simulations demonstrate the effectiveness of our solution in a heterogeneous environment, where users may potentially apply a number of different RAT selection procedures.

The Hough transform line detection method has been established as a viable technique for track-before-detect (TBD). However, its basic operation of binning and accumulating votes in the parameter space is computationally expensive. A more critical weakness of Hough transform is its dependence on parameter tuning (e.g., bin size and various thresholds), which can be non-intuitive and data-dependent. This leads to low detection rates in data with low signal-to-noise ratio and significant clutter. In this paper we propose a line detection algorithm with guaranteed global optimality for TBD. Our algorithm is based on the plane sweep algorithm for robust linear regression, with novel modifications to ensure its applicability under the TBD setting. Unlike the Hough transform, our algorithm has only one parameter to set (essentially the sensor false alarm rate) and can deterministically find the best solution according to a well-defined criterion. Simulation results on multi-dimensional TBD problems validate the accuracy and efficiency of our method.

This paper considers the exploitation of temporal correlation in incident sources in a narrowband array processing scenario. The MLE and CRB are derived for parameter estimation of spatially uncorrelated first order Gaussian autoregressive source signals with additive Gaussian spatially and temporally uncorrelated sensor noise. These are compared to the MLE and CRB for the usual uncorrelated (WN) sources model. The paper deals with the case where the number of data snapshots is small. Numerical simulations show that (i) there is no significant performance gain in the correlated signal case, and significantly, (ii) the WN MLE performance does degrade in the presence of source correlation, which appears to be in contrast to some recently published work.

We consider the prediction of a target's destination and simultaneously recover its filtered trajectory. Two novel models for trajectories with known destinations are presented using reciprocal stochastic processes and stochastic context-free grammars. We present a destination-aware syntactic tracker which uses conventional state-space estimates from a legacy tracker to perform prediction and trajectory estimation. We also provide statistical signal processing algorithms for model prediction and maximum likelihood sequence estimation using the proposed trajectory models. Simulation results show that both models considered in the paper have superior estimation performance compared to conventional hidden Markov modeling and can reliably predict the target's destination.

The high cost associated with the rollout of 3G services encourages operators to share network infrastructure. Network sharing poses a new challenge in devising fair and Pareto optimal resource allocation strategies to distribute system resources among users and operators in the network. Cooperative game theory provides a framework for formulating such strategies. In this paper, we propose two models (i.e. symmetric and asymmetric) for cooperative resource bargaining among the users and mobile virtual network operators (MVNOs) of each operator in shared networks based on the concept of preference functions. The bargaining solutions proposed vary according to a parameter beta that considers the tradeoff between one's gain and the losses of others. The well-known Nash and Raiffa- Kalai-Smorodinsky solutions are special instances of the solutions proposed. The symmetric model assumes that all players have equal bargaining powers while in the asymmetric case, players are allowed to submit bids to the network operator to influence the final bargaining outcome. Due to the diversity of demand patterns, temporary resource exchange among operators can provide benefits in terms of better communication quality to their users. To avoid selfish behaviour of the operators, we propose a resource sharing model that allocates extra resources based on the past allocations and contributions of each operator.

This letter proposes a generalized method to construct complex wavelets under the framework of rational multiresolution analysis, MRA(M), where M is a rational number. Theorems and examples are given for the construction of complex rational orthogonal wavelets (CROWs) whose real and imaginary parts form an exact Hilbert transform pair. Since the classical Mallat's MRA is a special case of the rational MRA(M) with M=2, the theorems hold for the construction of complex dyadic wavelets. Based on a rational MRA(M) with 1<M<2, the constructed CROWs not only achieve the benefit by capturing the phase information that the real-valued wavelets are lacking but also have the unique fine rational orthogonal property that is suited for specific application scenarios where binary orthogonality achieved by dyadic wavelets is not sufficient for the scale resolution. The CROWs' application in communications as the modulation signal pulse for PSK/QAM signaling is discussed. Specific communication scenarios that could benefit from the properties of this class of complex wavelets include the communication through a multipath/Doppler channel and new CROW-based multicarrier modulation (MCM)/orthogonal frequency division multiplexing (OFDM) systems

In this paper we investigate the performance of 802.11 carrier sense multiple access with collision avoidance algorithm for infrastructure networks under finite load conditions. An analytical model based on discrete time Markov chain is proposed to compute essential performance characteristics. The closed form solutions for the chain's stationary distribution and system throughput are derived. We model each wireless station as an M/G/1/ queue. The model accounts for arbitrary packet arrival rate, wireless channel access delay, transmission retry limits and arbitrary packet size distribution. Analytical results are verified through extensive simulation.

This letter addresses the issue of joint space-time trellis decoding and channel estimation in time-varying fading channels that are spatially and temporally correlated. A recursive space-time receiver which incorporates per-survivor processing (PSP) and Kalman filtering into the Viterbi algorithm is proposed. This approach generalizes existing work to the correlated fading channel case. The channel time-evolution is modeled by a multichannel autoregressive process, and a bank of Kalman filters is used to track the channel variations. Computer simulation results show that a performance close to the maximum likelihood receiver with perfect channel state information (CSI) can be obtained. The effects of the spatial correlation on the performance of a receiver that assumes independent fading channels are examined.

This paper presents the improvement of the spurious free dynamic range (SFDR) for digitisation using antenna arrays. Nonlinearities in the analogue-to-digital conversion degrade the overall SFDR of the digitisation process. We show that array processing, ie, the use of multiple sensors such as antennas or hydrophones with appropriate signal processing can improve the resulting SFDR at the beamform output. By taking advantage of spatial-temporal aliasing and the invisible regions, significant improvements can be obtained using linear, or more effectively, optimal beamforming.

This paper addresses the design and performance of time-recursive receivers for diversity based communication systems with flat Rayleigh or Ricean fading. The paper introduces a general state-space model for such systems, where there is temporal correlation in the channel gain. Such an approach encompasses a wide range of diversity systems such as spatial diversity, frequency diversity, and code diversity systems which are used in practice. The paper describes a number of noncoherent receiver structures derived from both sequence and a posteriori probability-based cost functions and compares their performance using an orthogonal frequency-division multiplex example. In this example, the paper shows how a standard physical delay-Doppler scattering channel model can be approximated by the proposed state-space model. The simulations show that significant performance gains can be made by exploiting temporal, as well as diversity channel correlations. The paper argues that such time-recursive receivers offer some advantages over block processing schemes such as computational and memory requirement reductions and the easier incorporation of adaptivity in the receiver structures.

The influence of a point spectrum on large sample statistics of the autoregressive (AR) spectral estimator is addressed. In particular, the asymptotic distributions of the AR coefficients, the innovations variance, and the spectral density estimator of a finite-order AR(p) model to a mixed spectrum process are presented. Various asymptotic results regarding AR modeling of a regular process with a continuous spectrum are arrived at as special cases of the results for the mixed spectrum setting. Finally, numerical simulations are performed to verify the analytical results.

Nonlinear intersymbol interference is often present in communication and digital storage channels. Decision-feedback equalizers (DFEs) can decrease this nonlinear effect by including appropriate nonlinear feedback filters. Although various applications of these types of equalizers have been published in the literature, the analysis of their stability and error recovery has not appeared. We consider a DFE with a nonlinear feedback filter based on a discrete Volterra series. We extend error propagation, error probability, stability, and error recovery time results for Nth order nonlinear channels.

This paper gives sufficient conditions for specifying the optimal linear filters for a hidden Markov process (HMP) and compares its performance with the optimal (i.e., minimum conditional variance) filter derived from the corresponding hidden Markov model using a simulation. The optimal filter performs much better at high signal-to-noise ratio (SNR) but the performance loss using the linear filter reduces as the SNR decreases.

A hidden Markov model method for estimating an a posterior distribution of the amplitude of QAM communications signals is presented. As the signal to noise ratio decreases the hidden Markov model method is shown to perform significantly better than a conventional histogram method for characterising the amplitude distribution. The HMM estimation is performed within a expectation maximisation method in order to improve the estimates of the transition probabilities used in the HMM and the resulting estimated amplitude distribution.

This paper develops a reduced-complexity online state sequence and parameter estimator for superimposed convolutional coded signals. Joint state sequence and parameter estimation is achieved by iteratively estimating the state sequence via a variable reduced-complexity Viterbi algorithm (VRCVA) and the model parameters via a recursive expectation maximization (EM) approach. The VRCVA is developed from a fixed reduced-complexity Viterbi algorithm (FRCVA). The FRCVA is a special case of the delayed decision-feedback sequence estimation (DDFSE) algorithm. The performance of online versions of the FRCVA, VRCVA, and the standard Viterbi algorithm (VA) are compared when they are used to estimate the state sequence as part of the reduced-complexity online state sequence and parameter estimator.

In this paper, a method for simultaneously demodulating and estimating the parameters of a number of convolutional coded communication signals incident on an antenna array is presented. The method has the potential to increase the throughput of current multiple-access channel systems, e.g., satellite communications and digital mobile cellular phones, by using an antenna array. The contribution of this paper is the use of sequence estimation combined jointly with parameter estimation in array processing problems. A hidden Markov-model-based technique, the segmental k-means algorithm, is applied to the problem. This algorithm is an iterative procedure with two steps per iteration. The first step involves computing the most likely state sequence for each of the signals (demodulating the signals) given estimates of the signals' parameters. The second step refines the parameter estimates using the signals' mostly likely state sequence estimates. In the simulations presented, it is shown that a significant improvement in the accuracy of the demodulated signals and in the estimation of the signals' angle of arrivals is obtained when compared to a deterministic maximum likelihood estimation method.

The paper addresses the problem of blind equalization of constant modulus signals which are degraded by frequency selective multipath propagation and additive white noise. An adaptive observer is used to update the weights of an FIR equalizer in order to restore the signal's constant modulus property. The observer gain is selected using fake algebraic Riccati methods in order to guarantee local stability. The performance of this method is compared to the constant modulus algorithm for simulated FM-FDM signals and exhibits significantly better convergence properties, particularly for heavy-tailed noise.

A method of jointly estimating the parameters of a number of superimposed convolutional coded communication signals incident on an antenna array and demodulating these signals is presented. The method would allow simpler arrays to be designed due to the threshold extension obtained by this method. It also has the potential to increase the throughput of current multiple access channel systems, for example, satellite communications and digital mobile cellular phones, by using an antenna array. The concept is based on the use of sequence estimation combined jointly with parameter estimation in array processing problems. In the simulations it is shown that a significant improvement in the accuracy of the demodulated signals and in the estimation of the signals angle of arrivals is obtained compared to a deterministic maximum likelihood estimation method.

This paper proposes a robust detection statistic for signals whose parameters are uncertain. Standard detection schemes generally use time domain correlation which can be related to correlation based on the Wigner-Ville distribution by Moyal's identity. This paper shows that a more robust detection statistic is achieved by using generalised patterns in time-frequency space and deriving a non-linear time domain correlation. The performance of the robust detection statistic is evaluated with the aid of receiver operating curves, two robust detection examples are given.

The problem of period uncertainty when evaluating spectrum estimates for wide sense cyclostationary processes is addressed in this paper. In particular, the extended Kalman filter (EKF) and a parallel bank of Kalman filters are investigated as different methods for adaptive estimation of a time-varying period. An example is given concerning an AR(1) process and a number of time-varying periods are adaptively tracked for different periodic functions. Convergence characteristics are also assessed. Finally, a combined detection-estimation approach is also investigated.

A method of determining the constraint length and generating polynomials of rate 1/L convolutional coded data using only the encoded data is presented.>

The use of the extended Kalman filter (EKF) is heavily entrenched in non-linear signal processing applications. However linearisation errors inherent in the specification of an EKF can severely degrade its performance. The paper presents a new approach to the robustification of the EKF by application of robust linear design methods based on the H/sub /spl infin// norm minimisation criterion. The results of simulations are presented to demonstrate an advantage for the demodulation of frequency and phase modulated signals.>

The paper addresses the problem of determining which one of a finite number of nonlinear dynamic systems generated a given noisy measured signal. An approximate likelihood ratio test (LRT) is proposed which consists of bank of an extended Kalman filters (EKFs) each tuned to one of the candidate signal models. The prediction error sequences of each EKF are used to form the LRT since each is nominally approximately zero-mean Gaussian with known covariance if it matches the measured signal. The Gaussian approximation is good at high signal-to-noise ratios (SNRs) but can degrade rapidly as the SNR decreases. The paper proposes a robustification of the test which is based on Huber's (1965) robust LRTs.>

This work addresses the problem of trying to take advantage of the nominally periodic nature of a diesel engine running at constant speed to obtain a time-varying spectral description of vibration data over the shaft period. It requires addressing a number of issues, most notably period estimation and removal of tonal components whose presence in a time-varying spectrum is redundant and undesirable. Use of a recently designed method to identify sinusoids, in conjunction with an adaptive tracking algorithm, results in significant improvement in simulations, and reasonably good improvement in the case of the diesel vibration data, especially when considering the complexity of the stochastic structure of this data.>

Two problems of interest in wideband signal processing are addressed. The first problem concerns the synthesis of wideband radar/sonar signal waveforms given the desired shape (as a function of time delay and time scale) of the cross correlator (matched filter) output; this function is termed the wideband ambiguity function (WBAF). Constraints on the transmitted waveforms that are manifest as closed convex subsets in the delay-scale space can also be incorporated into the synthesis procedure. The alternating convex projection theorem is used to generate a (weakly) convergent iterative algorithm for waveform synthesis. The second problem addressed concerns the closely related problem of signal recovery from R.A. Altes' (IEEE Trans. Acoust., Speech, Signal Proc., vol.38, no.6, p.1005-12, June 1990) Q-distribution and modified versions thereof. The Q-distribution is intimately related to the WBAF via a unitary transformation. Admissible modifications include affine convolutions and other operators with closed convex range. The application in terms of filtering in a time-scale domain is considered.>

A theory of symmetric phase shift keying is developed using quantum mechanics. The minimum average probability of error is then determined for phase-shift-keyed optical communication systems which use physical states. It is shown that this minimum error is achieved by using states which are represented in the number state basis by discrete prolate spheroidal sequences.>

The author addresses the problem of inverting a general member of the class of bilinear joint time-frequency signal representations introduced by L. Cohen (1966), i.e. obtaining the signal generating a given representation. Necessary and sufficient conditions for the exact inversion are introduced and are shown to be too restrictive to be practically useful. A constrained least-squares approach is formulated, and a solution is given by application of the method of alternating convex projections. The author also considers regularization of the method and examines some extensions of the approach.>

The problem of estimating the instantaneous frequency (IF) of a discrete time FM signal in additive white noise is addressed. The IF laws are modeled as first-order stationary Markov processes defined on the unit circle and a recursive algorithm for determining the maximum a posteriori probability IF sequences is derived. The basic single signal algorithm is of complexity O(N/sup 3/T) where T is the length of the signal segment and N is the number of subintervals of discretization of (0,2 pi ) used for the maximization step. Some examples are given to illustrate the tracking ability of the method.>

An automatic time-varying filter design technique is presented, based on windowing in the time-frequency plane about the IF of the signal under analysis, followed by signal synthesis using the Wigner-Ville distribution. The choice of the IF estimation algorithm determines the performance of the method. Algorithms for IF estimation are presented, and their performance in additive noise is discussed. The method is applicable to automatic filtering of FM signals. A review is presented of the various approaches followed by the authors and others.>

Consideration is given to the generalization of stationary cross spectral analysis methods to a class of nonstationary processes, specifically, the class of semistationary finite energy processes possessing sample functions that are of finite energy almost surely. A new quantity called the time-frequency coherence (TFC) is defined, and it is demonstrated that its properties are analogous to those possessed by the stationary coherence function. The problem of estimating the TFC by using elements of L. Cohen's (1966) class of joint time-frequency representations is investigated. It is shown that the only admissible estimators are those based on the class of time-frequency smoothed periodograms. Thus, the familiar procedure of segmentation and (smoothed) short-time Fourier analysis cannot be improved upon (within the framework considered) by the use of the higher-resolution nonparametric time-frequency methods. Procedures for selection of the appropriate estimators and a possible application are suggested.>

The author applies ideas from tomographic imaging to the problem of enhancing the resolution of time-frequency spectrum estimators based on elements of Cohen's (1966) class of time-frequency distributions. A multiple-least-squares (L/sub 2/) deconvolution method is introduced, and reblurring is used to stabilize the method. Multiple window functions are used to simulate tomographic slices. The estimated posterior density weighted (EPDW) algorithm from emission tomography is also applied to this problem, and the results are compared for some simulated signals. It is shown that the EPDW algorithm is superior, offering a higher degree of noise robustness and resolution enhancement.>

The Wigner-Ville Distribution (WVD) has recently been shown to be a valuable tool for time-Frequency Signal Analysis. Its first order moments yield directly the instantaneous frequency of the Signal. The problem of applying a window to the WVD is shown to be the same as that occurring when one uses the Fourier Transform. The choice of any classical window does not affect the estimation of the instantaneous frequency fi(t) in both deterministic and random cases. It is shown that the WVD estimator of the "evolutive spectrum" is unbiased in both time and frequency. Accuracy and statistical stability of the method are discussed. Its application to the analysis of microstructure temperature gradient signals shows that the WVD exhibits more information about the turbulence effect than the Short Time Fourier Transform.