Search

Compress-forward (CF) relays can improve communication rates even when the relay cannot decode the source signal. Efficient implementation of CF is a topic of contemporary interest, in part because of its potential impact on wireless technologies such as cloud-RAN. There exists a gap between the performance of CF implementations in the high spectral efficiency regime and the corresponding information theoretic achievable rates. We begin by re-framing a dilemma causing this gap, and propose an approach for its mitigation. We utilize trellis coded quantization (TCQ) at the relay together with multi-level coding at the source and relay, in a manner that facilitates the calculation of bit LLRs at the destination for joint decoding. The contributions of this work include designing TCQ for end-to-end relay performance, since a distortion-minimizing TCQ is suboptimum. The reported improvements include a 1dB gain over prior results for PSK modulation.

Recent progress in the understanding of the behavior of the interference channel has led to valuable insights: first, discrete signaling has been discovered to have tangible benefits in the presence of interference, especially when one does not wish to decode the interfering signal, i.e., the interference is treated as noise, and second, the capacity of the interference channel as a function of the interference link gains is now understood to be highly irregular, i.e., non-monotonic and discontinuous. This work addresses these two issues in an integrated and interdisciplinary manner: it utilizes discrete signaling to approach the capacity of the interference channel by developing lower bounds on the mutual information under discrete modulation and treating interference as noise, subject to an outage set, and addresses the issue of sensitivity to link gains with a liquid metal reconfigurable antenna to avoid the aforementioned outage sets. Simulations illustrate the effectiveness of our approach.

This paper investigates basic trade-offs between energy and delay in wireless communication systems using finite blocklength theory. We first assume that data arrive in constant stream of bits, which are put into packets and transmitted over a communications link. Our results show that depending on exactly how energy is measured, in general energy depends on $\sqrt {d^{-1}}$ or $\sqrt {d^{-1}\log {d}}$ , where $d$ is the delay. This means that the energy decreases quite slowly with increasing delay. Furthermore, to approach the absolute minimum of −1.59 dB on energy, bandwidth has to increase very rapidly, much more than what is predicted by infinite blocklength theory. We then consider the scenario when data arrive stochastically in packets and can be queued. We devise a scheduling algorithm based on finite blocklength theory and develop bounds for the energy-delay performance. Our results again show that the energy decreases quite slowly with increasing delay.

This paper studies the low-SNR regime performance of a scalar complex K -user interference channel with Gaussian noise. The finite bandwidth case is considered, where the low-SNR regime is approached by letting the input power go to zero while bandwidth is small and fixed. We show that for all \delta>0 there exists a set with non-zero measure (probability) in which the wideband slope per user satisfies Slope, Comment: submitted to Information Theory, IEEE Transactions on

This paper proposes a new method for impulsive source separation (deinterleaving) in multi-input single-output systems based purely on impulse spacing. The impulse spacing for each source is modeled as a renewal process. To classify sources, an algorithm is developed to maximize the likelihood of impulse assignment based on the impulse spacing distributions. A theoretical lower bound on performance is derived and the algorithm is shown to get reasonably close to the bound. The algorithm is extended for use when the impulse spacing distribution parameters are unknown. Performance is evaluated on multiple application examples.

A classic application of description length is for model selection with the minimum description length (MDL) principle. The focus of this paper is to extend description length for data analysis beyond simple model selection and sequences of scalars. More specifically, we extend the description length for data analysis in Gaussian graphical models. These are powerful tools to model interactions among variables in a sequence of i.i.d Gaussian data in the form of a graph. Our method uses universal graph coding methods to accurately account for model complexity, and therefore provide a more rigorous approach for graph model selection. The developed method is tested with synthetic and electrocardiogram (ECG) data to find the graph model and anomaly in Gaussian graphical models. The experiments show that our method gives better performance compared to commonly used methods., Comment: Submitted to ISIT 2021

This paper asks a basic question: how much training is required to beat a universal source coder? Traditionally, there have been two types of source coders: fixed, optimum coders such as Huffman coders; and universal source coders, such as Lempel-Ziv The paper considers a third type of source coders: learned coders. These are coders that are trained on data of a particular type, and then used to encode new data of that type. This is a type of coder that has recently become very popular for (lossy) image and video coding. The paper consider two criteria for performance of learned coders: the average performance over training data, and a guaranteed performance over all training except for some error probability $P_e$. In both cases the coders are evaluated with respect to redundancy. The paper considers the IID binary case and binary Markov chains. In both cases it is shown that the amount of training data required is very moderate: to code sequences of length $l$ the amount of training data required to beat a universal source coder is $m=K\frac{l}{\log l}$, where the constant in front depends the case considered., Comment: Submitted to IEEE Transactions on Information Theory

We investigate the performance of discrete (coded) modulations in the full-duplex compress-and-forward relay channel using multilevel coding. We numerically analyze the rates assigned to component binary codes of all levels. LDPC codes are used as the component binary codes to provide error protection. The compression at the relay is done via a scalar quantizer whose output is mapped to a codeword through LDPC codes. A compound Tanner graphical model and information exchange algorithm are described for the joint decoding of both messages sent from the source and relay. Simulation results show that the performance of the proposed system based on multilevel coding is better than that based on BICM, and is separated from the SNR threshold of the known compress-and-forward achievable rate by two factors consisting approximately of the sum of the shaping gain (due to the scalar quantization) and the separation of the LDPC code implementation from AWGN capacity.

The aim of using atypicality is to extract small, rare, unusual and interesting pieces out of big data. This complements statistics about typical data to give insight into data. In order to find such &ldquo, interesting&rdquo, parts of data, universal approaches are required, since it is not known in advance what we are looking for. We therefore base the atypicality criterion on codelength. In a prior paper we developed the methodology for discrete-valued data, and the current paper extends this to real-valued data. This is done by using minimum description length (MDL). We develop the information-theoretic methodology for a number of &ldquo, universal&rdquo, signal processing models, and finally apply them to recorded hydrophone data and heart rate variability (HRV) signal.

In [1] , the affiliation of A. Host-Madsen was incorrect. The affiliation should read: A. Host-Madsen is with the Department of Electrical Engineering, University of Hawaii at Manoa, Honolulu, HI 96822 USA (e-mail: ahm@hawaii.edu).

We pursue interference mitigation via the integration of two key ideas. First, understanding the behavior of available rates under discrete signaling which has recently been shown to be promising in the interference channel. This part calls for calculation of good bounds on post-interference mutual information under discrete signaling, as a function of the forward and cross channel gains. Second, the capacity of the interference channel is known to be very irregular, so for any target capacity there are “outage” sets that we aim to avoid by using reconfigurable antennas. For the first component, we report an analytical lower bound on the mutual information and establish a constant gap $O(\log\gamma)$ to capacity (excepting an outage set) that is derived using a purely discrete signaling. This result outperforms the gap reported by Dytso et al that was generated via a mixed discrete-continuous input strategy. In the second part of this work, we propose to use a reconfigurable antenna technology involving liquid metal antennas that can steer the channel away from the outage scenarios and therefore facilitate higher values of coded rates for the two-user symmetric Gaussian interference channel. The viability of the proposed technique is studied via simulations.

This paper has dual aims. First is to develop practical universal coding methods for unlabeled graphs. Second is to use these for graph anomaly detection. The paper develops two coding methods for unlabeled graphs: one based on the degree distribution, the second based on the triangle distribution. It is shown that these are efficient for different types of random graphs, and on real-world graphs. These coding methods is then used for detecting anomalous graphs, based on structure alone. It is shown that anomalous graphs can be detected with high probability., To be presented at ISIT'18

The deterministic-coding capacity of a network is the capacity of a network when nodes are restricted to transmitting reliable information, that is, (asymptotically) deterministic functions of the source messages. This is a generalization of network coding to wireless networks. The main contribution of this paper is to develop new outer bounds on the deterministic-coding capacity in the low-power regime, $ \mathrm {SNR}\to 0$ . This is accomplished by developing a technique for finding the limits of single letter bounds for $ \mathrm {SNR}\to 0$ even when these cannot be evaluated for finite SNR. It is shown that these outer bounds are tight in a number of cases, including the butterfly network of Ahlswede et al. in a wireless setting.

A central question in the era of 'big data' is what to do with the enormous amount of information. One possibility is to characterize it through statistics, e.g., averages, or classify it using machine learning, in order to understand the general structure of the overall data. The perspective in this paper is the opposite, namely that most of the value in the information in some applications is in the parts that deviate from the average, that are unusual, atypical. We define what we mean by 'atypical' in an axiomatic way as data that can be encoded with fewer bits in itself rather than using the code for the typical data. We show that this definition has good theoretical properties. We then develop an implementation based on universal source coding, and apply this to a number of real world data sets., 40 pages

With the enormous amount of data generated through the internet and sensors, Internet of Things, it becomes too overwhelming for humans to examine it all. One solution is to reduce the data to a set of statistics. The perspective in this paper is the opposite, namely that most of this data is just background noise, and the interesting parts are those that deviate from background noise, the parts that are atypical. In order to find such “interesting” parts of data, universal approaches are required, since it is not known in advance what we are looking for. Our approach is to use Rissanen's minimum description length (MDL) as a tool for that. We would like to be able to find both short and long atypical sequences of data, and we therefore need accurate expressions of MDL, without prior assumptions. In this paper we develop a modified predictive MDL method that works better for short sequences.

This paper analyzes the performance of discrete input distributions (coded modulation) in interference channels. This approach is motivated in part by the necessity of using coded modulation in practical systems, and in part by the potential of discrete distributions for interference alignment as well as the importance demonstrated by Dytso et al. of discrete input distributions for transmission over the $2\times 2$ interference channel when treating interference as noise. The contribution of this work includes the establishment of achievable rates subject to discrete modulations. In the process, new bounds involving the minimum distance of the sum of discrete modulations have been developed that are useful for facilitating further work in this area. These bounds are then used for finding achievable rates for $3\times 3$ interference channels. It is also shown that interference alignment can be efficiently employed with discrete modulations at finite SNR even with imperfect channel state information at the transmitter.

With the exponential growth in information, “Big Data,” a key question is what to do with this information. A classic possibility is to characterize it through statistics. The perspective in this paper is the opposite, namely that most of the value in the information is in the parts that deviates from the average, that are unusual, atypical. Think of art: The valuable paintings or writings are those that deviate from the norms, that are atypical. With the same perspective think of social networks: Flickr is an example that has a rating for images based on how interesting they are. In previous works, researchers introduced algorithms to find the interesting images based on human psychological taste; but in this paper, we are proposing a new method to rate an image based on their level of interestingness using atypicality, in which interestingness is equivalent to having atypical structure. The paper first discusses what exactly should be understood by “atypical.” It has to be a well defined theoretical concept corresponding to some intuitive idea of atypicality, which when applied gives useful results. This is followed by applying the notion of atypicality to the general discrete scenario with finite alphabet and then putting it in to the framework of the quantized wavelet image coding. Finally our proposed algorithm is applied on Labeled Faces in the Wild (LFW) database with 13,233 images in order to find the interesting ones.

With the enormous amount of data generated through the internet and sensors, Internet of Things, it becomes too overwhelming for humans to examine it all. One solution is to reduce the data to a set of statistics. The perspective in this paper is the opposite, namely that most of this data is just background noise, and the interesting parts are those that deviate from background noise, the parts that are atypical. In order to find such “interesting” parts of data, universal approaches are required, since it is not known in advance what we are looking for. The paper develops new algorithms for detecting atypical data based on information theory concepts. These are applied to a number of real-world data.

Atypicality is a new concept that uses a codelength-based deviation from the norm to find the interesting rare events. In a previous paper we have developed an information theoretic approach for discrete data. Then in the two other papers, we came up with an extension to the real-valued models for Gaussian and vector Gaussian cases. In the current paper we generalize our real-valued model to the class of exponential family. We include a comprehensive table that summarize the approximated atypicality criterion and bounds for common distributions in the exponential family model.

Vector Orthogonal Frequency Division Multiplexing (OFDM) for single transmit antenna systems is a general transmission scheme, where OFDM and Single-Carrier Frequency Domain Equalization (SC-FDE) can be treated as two special/extreme cases. Due to its flexibility, it has drawn more and more attention recently. So far, all the studies about Vector OFDM assume the Maximum Likelihood (ML) receiver. In this paper, we investigate the performance of Vector OFDM with linear receivers, i.e., the Zero-Forcing (ZF) and Minimum Mean Square Error (MMSE) receivers. We first show that the detection SNR gap between the MMSE and ZF receivers increases with both channel SNR and the vector blocks (VB) size defined in Vector OFDM. Then, it is proved that for both ZF and MMSE receivers, all the transmitted symbols have equal performance. This is different from the Vector OFDM with ML receiver, where different VBs may have different coding gain, and thus may have different performances. We analyze the diversity order for Vector OFDM with MMSE receiver, and show that, regardless of the Vector OFDM symbol length N, the diversity order can be represented as min{[M2-R],D}+1, where M is the VB size, R is the spectrum efficiency in bits/symbol, and D is the maximum delay of the multipath channel. For Vector OFDM with ZF receiver, we show that the diversity order equals 1 and the performance is the same as the conventional OFDM at high SNR.

This paper derives a formula for the wideband slope for outage capacity and consider its application to some specific wireless channels. For the broadcast channel, the formula indicates that superposition is always superior to time-division multiple access (TDMA). On the other hand, for the interference channel, we show that TDMA is better than superposition for most realistic situations.

Atypical sequences are subsequences of long sequences that deviate from the ‘normal’ data. In previous papers we have developed an information-theoretic approach to such sequences for discrete and real-valued data. In the current paper we extend the principle of real-valued data that follows vector Gaussian models, which allows for finding relationship between data. We include a simple application to stock market data.

In this paper, we consider cooperation in the low power (low SNR) regime for the multiple access channel with the assumption that the transmitters have no channel state information. A relevant performance measure to consider is therefore the outage capacity. We develop cooperation methods based on multiplexed coding in conjunction with rateless codes and find the achievable rates and in particular the minimum energy per bit required to achieve a certain outage probability. We consider two modes of operation: full duplex [code-division multiple access (CDMA)], where nodes can transmit and receive simultaneously on the same frequency band, and half duplex [frequency-division multiple access (FDMA)], where the nodes transmit and listen on different frequency bands. We show that, perhaps surprisingly, there is little loss in performance when using FDMA. Furthermore, our results show that multiplexed rateless codes come within 0.1 dB of the outer bound on capacity. We also develop practical rateless coding methods for FDMA using multiplexed Raptor codes which operate within 0.52 and 1.1 dB of the theoretical limit for the two- and four-user case, respectively.

An investigation of heart rate variability (HRV) and respiratory sinus arrhythmia (RSA) indices using data obtained from Doppler radar cardiopulmonary remote sensing is presented in this paper. High accuracy in extracting the HRV and RSA indices was achieved using a direct-conversion quadrature radar system with linear demodulation method. This method was validated using data obtained from 12 human subjects in seated and supine positions. For supine position measurements, all standard deviation of normal beat-to-beat interval indices from Doppler radar and electrocardiogram reference differed less than plusmn9 ms of each other, while all the root mean square of differences of successive normal beat-to-beat interval indices differed less than plusmn76 ms. The measurements from subjects in seated and supine positions with normal RSA demonstrated that the results from radar correlated well with both respiratory piezoresistor chest belts. Higher level of HRV and RSA was detected in seated position, as expected.

We consider the design of cooperation coding schemes for a two-user multiaccess channel (MAC). In particular, we consider two block Markov coding schemes, namely, the multiplexed coding and the superposition coding. Although the fully multiplexed (FMP) coding outperforms the superposition coding in theory, it is difficult to implement with practical error-correction codes. We, therefore, introduce a partially multiplexed (PMP) coding scheme for code rate R < 0.5, and propose a simplified superposition coding scheme. The outage analysis shows that these cooperation schemes provide significant gains over the non- cooperative MAC in the low-power regime when the bandwidth W rarr infin. We then propose a PMP code design using irregular repeat accumulate (IRA) codes. The design method can be extended to build the low-rate PMP codes using irregular repeat zigzag Hadamard (IRZH) codes, a new class of capacity achieving low-rate codes. Simulation results demonstrate that the block Markov multiplexed coding using PMP-IRA codes outperforms the practical superposition coding by 2 dB at 10-2 FER, and it provides a 10-dB gain over the noncooperative MAC. The multiplex coding using PMP-IRZH codes offers 1-dB gain over the superposition coding, and a 9.5-dB gain over noncooperative MAC at 10-2 FER. Moreover, the PMP-IRZH coding performs only 0.5 dB away from the outage capacity of multiplexed coding in the low-power regime.

We study transmit power adaptation and capacity-approaching coding/decoding in multiple-input-multiple-output (MIMO) Rayleigh fading channels under the assumption that perfect channel state information (CSI) is known at both the transmitter and the receiver. The capacity in MIMO systems with transmitter CSI can be achieved via two schemes: 1) a multiple-coding scheme with temporal and spatial water filling and 2) a single-coding scheme with temporal and spatial water filling. The former requires an infinite number of different codes, and the latter one requires interblock coding and, therefore, a very long code. We propose three simple but powerful methods for transforming the MIMO fading channel into a set of additive white Gaussian noise channels to which standard codes for the Gaussian channel can then be applied. We show through a number of examples that these methods can closely approach channel capacity. As a by-product, the closed-form expressions of the marginal probability density function of the ordered and unordered eigenvalues of Wishart matrices are derived.

In this paper, we compare two classes of linear interference suppression techniques for downlink TDD-CDMA systems, namely, linear multiuser detection methods (receiver processing) and linear precoding methods (transmitter processing). For the linear precoding schemes, we assume that the channel state information (CSI) is available only at the transmitter but not at the receiver (i.e., ultra simple receivers). We propose several precoding techniques and the corresponding power control algorithms. The performance metric used in the comparisons is the total power required at the transmitter to achieve a target SINR at the receiver. Our results reveal that in general multiuser detection and precoding offer similar performance; but in certain scenarios (e.g, low BER requirements or use of random spreading sequences), precoding can bring a substantial performance improvement. These results motivate the use of precoding techniques to reduce the complexity of the system and the mobile terminals (only a matched-filter to the own spreading sequence is required without CSI). Moreover, it is shown that the proposed chip-wise linear MMSE precoding method is optimal in the sense that it requires the minimum total transmitted power to meet a certain receiver SINR performance.

This paper considers the problem of continuously streaming bits over a AWGN channel, a simple model of video or audio streaming. The bits are required to be decoded with a maximum delay, a deadline, of d. The question the paper seeks to answer is: what is the fundamental cost of the deadline in terms of energy per bit Eb? The paper answers this question for a number of cases, depending on the bandwidth available for transmission. A common conclusion is that for an excess energy of ΔEb, i.e., the energy above that required in the limit d → ∞, the required delay is d ∼ ΔEb−2 which is much worse than the linear relationship between bandwidth and Eb−1.

Atypical sequences are subsequences of long sequences that deviates from the ‘normal’ data. In a previous paper we have developed an information theory approach to such sequences for discrete data. In the current paper we extend this principle to real-valued data, whereby it is possible to use signal processing tools to search for atypical data. The application of this principle is to extract a few interesting sets of information from ‘big data’ sets. We include a simple application to stock market data.

In this work, we investigate the minimum power of transmitting Gaussian multiterminal sources over the Gaussian multiple access channel (MAC) with bandwidth expansion. Distributed encoders observe the individual components of the multiterminal sources and attempt to describe them to a central decoder. Instead of assuming the sources and the channel have matched bandwidth, we consider the scenario when the transmission is subject to a bandwidth constraint in terms of the number of channel uses per source sample. We study the minimum transmission power such that the decoder can reconstruct the individual sources and meet the mean squared error (MSE) constraints on them. For any given bandwidth and distortion constraints, we optimize uncoded transmission to obtain an upper bound on the minimum transmission power. A hybrid coding scheme is then proposed to obtain a tighter upper bound.

Cooperative diversity is a novel technique proposed for conveying information in wireless ad hoc networks, where closely located single-antenna network nodes cooperatively transmit and/or receive by forming virtual antenna arrays. For its building blocks, the relay channel and the two-transmitter, two-receiver cooperative channel, we survey the latest advances made in determining the theoretical capacity bounds and describe the best practical code designs reported so far. Both theory and practice predict that cooperative communication can provide increased capacity and power savings in ad hoc networks.

Motivated by the result in [12] we present a new block iterative multiuser detection algorithm utilizing the constant modulus (CM) property of data symbols. The proposed iterative detector is developed based on a constrained least square (LS) cost function and CM projection. It can be seen as an extension of least squares constant modulus algorithm [1] in a sense that it minimizes the CM and the LS cost functions iteratively. Simulation study shows that even for a reasonably small number of symbols, say around 100 received symbols, the proposed iterative detector approaches the performance of ideal MMSE detector and considerably outperforms the conventional blind detectors, such as subspace blind method 171 or MOE detector [5].

In a cooperative diversity network, users cooperate to transmit each others' messages; to some extent nodes therefore collectively act as an antenna array and create a virtual or distributed multiple-input multiple-output (MIMO) system. In this paper, upper and lower bounds for the information-theoretic capacity of four-node ad hoc networks with two transmitters and two receivers using cooperative diversity are derived. One of the gains in a true MIMO system is a multiplexing gain in the high signal-to-noise ratio (SNR) regime, an extra factor in front of the log in the capacity expression. It is shown that cooperative diversity gives no such multiplexing gain, but it does give a high SNR additive gain, which is characterized in the paper

We study the capacity of multiple-input multiple- output (MIMO) relay channels. We first consider the Gaussian MIMO relay channel with fixed channel conditions, and derive upper bounds and lower bounds that can be obtained numerically by convex programming. We present algorithms to compute the bounds. Next, we generalize the study to the Rayleigh fading case. We find an upper bound and a lower bound on the ergodic capacity. It is somewhat surprising that the upper bound can meet the lower bound under certain regularity conditions (not necessarily degradedness), and therefore the capacity can be characterized exactly; previously this has been proven only for the degraded Gaussian relay channel. We investigate sufficient conditions for achieving the ergodic capacity; and in particular, for the case where all nodes have the same number of antennas, the capacity can be achieved under certain signal-to-noise ratio (SNR) conditions. Numerical results are also provided to illustrate the bounds on the ergodic capacity of the MIMO relay channel over Rayleigh fading. Finally, we present a potential application of the MIMO relay channel for cooperative communications in ad hoc networks.

The analytical performance of the subspace-based blind linear minimum mean-square error (MMSE) multiuser detection algorithm in general multipath multi-antenna code-division multiple-access (CDMA) systems is investigated. In blind multiuser detection, the linear MMSE detector of a given user is estimated from the received signals, based on the knowledge of only the spreading sequence of that user. Typically, the channel of that user must be estimated first, based on the orthogonality between the signal and noise subspaces. An asymptotic limit theorem for the estimate of the blind linear detector (when the received signal sample size is large) is obtained, based on which approximate expressions of the average output signal-to-inference plus noise ratios (SINRs) and bit error rates (BERs) for both binary phase-shift keying (BPSK) and quaternary phase-shift keying (QPSK) modulations are given. Corresponding results for group-blind multiuser detectors are also obtained. Examples are provided to demonstrate the excellent match between the theory developed in this paper and the simulation results.

In code division multiple access (CDMA) systems employing linear adaptive receivers, the detector is typically estimated directly from the received signals, based on some partial knowledge about the system, e.g., signature waveforms of one or several users. We derive the Cramer-Rao lower bounds on the covariances of the estimated linear detectors, under three different assumptions on the mechanism for estimating the detectors, namely, a) finite-alphabet-based (FA) blind detectors, b) constant-modulus-based (CM) blind detectors, and c) second-order-moments-based (SO) blind detectors. These bounds translate into the upper bounds on the achievable signal-to-interference-plus-noise ratio (SINR) by the corresponding adaptive receivers. The results are asymptotic in nature, either for high signal-to-noise ratio (SNR) or for large signal sample size. The effects of unknown multipath channels on these performance bounds are also addressed. Numerical results indicate that while the existing subspace blind or group-blind detectors perform close to the SINR bound for the SO detectors, the SINR bounds for the FA and CM detectors are significantly higher, which suggests potential avenues for developing more powerful adaptive detectors by exploiting more structural information from the system.

In blind (or group-blind) linear multiuser detection, the detector is estimated from the received signals, with the prior knowledge of only the signature waveform of the desired user (or the signature waveforms of some but not all users). The performance of a number of such estimated linear detectors, including the direct-matrix-inversion (DMI) blind linear minimum mean square error (MMSE) detector, the subspace blind linear MMSE detector, and the form-I and form-II group-blind linear hybrid detectors, are analyzed. Asymptotic limit theorems for each of the estimates of these detectors (when the signal sample size is large) are established, based on which approximate expressions for the average output signal-to-interference-plus-noise ratios (SINRs) and bit-error rates (BERs) are given. To gain insights on these analytical results, the performance of these detectors in an equicorrelated code-division multiple-acces (CDMA) system is compared. Examples are provided to demonstrate the excellent match between the theory developed here and the simulation results.

A nonlinear group-blind technique is developed for joint detection of some given users' data in a CDMA uplink environment with the presence of unknown interference. This method performs the so-called "slowest-descent search" over a likelihood function of the desired users, starting from the estimate closest to the unconstrained maximizer of the likelihood function, and along mutually orthogonal directions where this likelihood function drops to the slowest. Simulation results show that this new nonlinear technique offers substantial performance improvement over the previously proposed linear group-blind multiuser detectors with little attendant increase in computational complexity. The problem of group-blind multiuser detection in the presence of both unknown interference and impulsive ambient noise is also treated under the framework of slowest-descent search, with the aid of a novel subspace-based robust interference cancellation scheme. It is seen that this robust group-blind method significantly outperforms the robust blind multiuser detection scheme proposed previously.

This paper considers the problem of joint time delay and impulse response estimation and its applications to, for example, echo cancellation for hybrid echos in telephone lines. The paper suggests a new algorithm based on considering the minimum mean-square error output of an RLS-type filter. The RLS filter is run for different possible test delays, and a hypothesis test is applied for each test delay to test if the true delay is within some interval around the test delay. It is shown that the algorithm has very low complexity (about the same as FTF-type filters), and the performance is investigated theoretically and through simulations. The algorithm can be used for a very low complexity echo canceller with convergence time almost as short as the echo bulk delay.

The effects of sampling and quantization on frequency estimation for a single sinusoid are investigated. The Cramer-Rao bound for 1-bit quantization is derived and compared with the limit of infinite quantization. It is found that 1-bit quantization gives a slightly worse performance, however, with a dramatic increase of variance at certain frequencies. This can be avoided by using four times oversampling. The effect of sampling when using nonideal antialiasing lowpass filters is therefore investigated through derivation of the Cramer-Rao lower bounds. Finally, fast estimators for 1-bit quantization, in particular, correlation-based estimators, are derived, and their performance is investigated. The paper is concluded with simulation results for four times oversampled 1-bit quantization.

The common signal processing problem of estimating some nonrandom parameters of a signal in additive noise is considered. The problem investigated in this paper is under what conditions an efficient estimator exists, i.e., an unbiased estimator with a variance equal to the Cramer-Rao lower bound (CRB). It is well known that if the signal is linear or, more generally, affine in the parameters and the noise Gaussian, an efficient estimator does exist. This paper shows that under some conditions, this is the only case where an efficient estimator exists.

This paper analyzes combinations of the linear minimum mean square error (MMSE) detector and a nonlinear parallel interference canceller (PIC) for multiuser code-division multiple-access (CDMA) detection. Both the case where all users' codes are known at the receiver and the case where only some codes are known are considered. An upper bound and an approximate formula for the error probability are derived and verified through simulation. It is shown that the combined MMSE/PIC detector can have a considerable performance gain over the MMSE. It is also shown that while the MMSE detector is sensitive to large code cross-correlation values, the combined MMSE/PIC detector is robust to code cross correlations. Finally, use of the MMSE/PIC detector for cellular systems with both inter- and intra-cell interference is considered.

A mathematical treatment of particle size and velocity estimation from two channel laser anemometry measurements is considered. Cramer-Rao bounds for the general case are derived, and the corresponding maximum likelihood estimator (MLE) is analysed through computer simulations. Alternative estimators are derived and their performance is characterized and compared with the performance of the MLE. It is shown that if only the size of the particle is of interest, a simple correlation-based estimator provides accurate results. Accurate velocity estimation from correlations are also possible based on a prior estimate of the particle size. The results predicted by theory are illustrated by some numerical examples.

We propose an iterative tuning process of blind/group-blind detector based on constant modulus property of digitally modulated signal. By adopting the iterative tuning process after applying blind/group-blind detectors, significant gains are attained compared to the original blind/group-blind detectors. The signal to interference and noise ratio (SINR) approaches to ideal MMSE detector even with reasonably small number of data samples.

One characteristic of the information age is the exponential growth of information, and the ready availability of this information through networks, including the internet — “Big Data.” The question is what to do with this enormous amount of information. One possibility is to characterize it through statistics — think averages. The perspective in this paper is the opposite, namely that most of the value in the information is in the parts that deviate from the average, that are unusual, atypical. Think of art: the valuable paintings or writings are those that deviate from the norms, that are atypical. The same could be true for venture development and scientific research. The paper first discusses what exactly should be understood by “atypical.” This is by no means straightforward. It has to be a well defined theoretical concept corresponding to some intuitive idea of atypicality, which when applied gives useful results. This is followed by a simple example of iid binary sequences. This example is simple enough that complete algorithms can be developed and analyzed, which give insights into atypicality. We finally develop a more general algorithm based on the Context Tree Weighing algorithm and apply that to heart rate variability.

This paper proposes a new method for impulsive source separation in multi-input single-output (MISO) systems based on impulse spacing. The primary application of the method is the classification of click trains from multiple marine mammals. The impulse spacing can be modeled as a renewal process, so to classify sources, an algorithm is developed to maximize the likelihood of click assignment based on the impulse spacing distributions. A theoretical lower bound on the performance is derived and the algorithm shown to get reasonably close to the bound. Additionally, the algorithm is further extended for use when the impulse spacing distribution parameters are unknown. The algorithm is tested with recordings of sperm whales, and the separation is shown to be accurate.

In this work, we investigate the minimum energy of transmitting correlated sources over the Gaussian multiple-access channel (MAC). Compared to other works on joint source-channel coding, we consider the general scenario where the source and channel bandwidths are not naturally matched. In particular, we proposed the use of hybrid digital-analog coding over to improve the transmission energy efficiency. Different models of correlated sources are studied. We first consider lossless transmission of binary sources over the MAC. We then treat lossy transmission of Gaussian sources over the Gaussian MAC, including CEO sources and multiterminal sources. In all cases, we show that hybrid transmission achieves the best known energy efficiency., Under revision

In this paper, we initiate a theoretical investigation of spectral estimation of randomly sampled stochastic processes using interpolation. We develop a new method for spectral estimation of Poisson-sampled stochastic processes. Our method is based on exponential interpolation from the sampled process followed by resampling and usual discrete Fourier transform. We prove that this method — with appropriate corrections — is asymptotically unbiased and show how problems arising from aliasing in the resampling process can be overcome. We compare this method with other methods for spectral estimation of Poisson-sampled processes, and show that — in most cases — the new method is superior in accuracy and speed.