Position estimation (geolocation) of a direct-sequence, spread-spectrum radio emitter is particularly challenging because of low signal-to-noise ratio at the radio sensors. In this letter, we develop a maximum-likelihood approach to geolocation that utilizes knowledge of the spreading sequence. Simulation results are used to show that spreading sequence knowledge can provide significant gains in position accuracy.

In this letter, a multistage group detection (MSGD) approach to nonlinear equalization is proposed. At each stage, symbols are divided into groups, and maximum-likelihood detection (MLD) within the group is performed. The MLD process identifies a number of promising group values. It is constrained to only consider candidate symbol values identified in the previous stage. Groups are formed by combining groups from the previous stage. For example, symbols are detected one at a time, then two at a time, then four at a time, and so on. The last stage is a form of constrained maximum-likelihood sequence detection (MLSD). The overall demodulation process can be viewed as approximate MLSD, and results show that near-MLSD performance is possible with reasonable complexity. While the High Speed Downlink Packet Access (HSDPA) cellular system is used as an example, the approach can be applied other situations, including the Long Term Evolution (LTE) uplink and MIMO.

As 3G DS-CDMA cellular systems evolve to higher data rates, linear equalization is being used to address inter-symbol interference introduced by dispersive channels. When parametric designs are employed, elements of an impairment or data covariance matrix must be computed as part of equalizer weight calculation. The conventional approach is to express these elements in terms of path delays and path coefficients. This approach requires a significant amount of computation, and performance is sensitive to errors in the path delay estimates. In this paper, an alternative approach is developed, based on directly estimating net channel coefficients corresponding to an extended set of delays that includes the tap or finger delays of the equalizer. This avoids the need for path delay estimates, improving robustness. Also, the elements of the impairment or data covariance can be expressed in terms of efficiently computed convolutions of these net channel coefficients, even though these elements correspond to correlations of cyclostationary interference. Complexity estimates are used to show computational savings, and performance simulations are used to show gains in robustness.

Direct-sequence code-division multiple-access (DSCDMA) cellular systems, such as wideband CDMA (WCDMA), are limited in performance by interference. Linear equalization, such as generalized Rake (G-Rake) receivers and transversal chip equalizers, can theoretically provide significant gains in performance by suppressing interference. In this paper, an adaptive G-Rake receiver is developed, employing practical algorithms for finger placement and weight computation. Finger placement is determined by selecting delays from a candidate set using a "maximal-weight" criterion. Weight computation includes estimation of an impairment covariance matrix using a parametric approach. While described in the context of the downlink, these algorithms can be used in the uplink as well. Both single-antenna and dual-antenna receivers are considered. Link and system performance are evaluated for the downlink, showing significant gains in high-rate coverage.

High speed packet access (HSPA), the latest evolution of the wideband code division multiple access (WCDMA) standard, provides higher uplink data rates using higher-order modulation (16-QAM) and multicode transmission. Equalization is needed when the channel is dispersive. Despite gains from linear equalization, there is room for further improvement. In this paper, we consider two enhancements: block equalization with joint detection and a generalization of arbitration. Block equalization with joint detection, originally developed for non-spread systems, involves jointly detecting symbols from multiple symbol periods and using a feedback filter to remove interference from past symbol periods. Here we extend this idea to parallel symbol streams, add feedforward filtering, and introduce the notion of block linear equalization. Arbitration, also originally developed for nonspread systems, involves bidirectional decision feedback equalization and selection. Here we generalize the arbitration concept to other forms of equalization and improve the selection process using constrained maximum-likelihood sequence estimation (MLSE). These enhancements are then evaluated via simulation for the HSPA uplink, demonstrating their efficacy.

In this paper, we study the performance of transmit diversity schemes in a wideband code-division multiple-access (WCDMA) system. We consider a space-time transmit diversity (STTD) scheme using the Alamouti code and a simple switch transmit diversity (STD) scheme in which the transmitter selects instantaneously the best antenna for transmission. Receiver impairment is generally modeled as a colored, complex Gaussian process, which gives rise to temporal and spatial correlations. First, we extend the G-Rake formulation to process a STTD signal, accounting for the space-time code structure as well as the temporal and spatial correlations of the impairments. The performance of STTD and STD for a WCDMA system is then evaluated. We show that STTD improves the coverage of low-to-medium rate services significantly for users at cell edge compared to the scheme without transmit diversity. Such improvements are observed even when the channels already have diversity through multipath and multiple receive antennas. Thus, STTD is well suited for broadcast when feedback from mobiles is not available. For high-data-rate availability, however, using one feedback bit to allow the STD scheme to transmit only on the better antenna achieves significantly better performance compared to STTD. This is because sending the desired signal through one radio channel makes it easier for linear equalizers to restore orthogonality. For mobiles dominated by self-interference due to loss of orthogonality, better equalization results in much improved data rates.

The system benefits of using an advanced dual-branch diversity receiver at the mobile terminal are studied in a Wideband Code-Division Multiple-Access (WCDMA) cellular system. The receiver considered is an extension of the generalized Rake (G-Rake) receiver which provides a form of array processing. In addition to exploiting the temporal correlation in receiver impairments and the orthogonality between spreading codes, spatial correlation is also accounted for in the formulation of finger combining weights. We show that this dual-branch G-Rake receiver, even with a pessimistic antenna configuration, has a potential to increase the system capacity, relative to the conventional single-branch Rake receiver, by 94% for voice communications. For data communications, high-data-rate availability can be significantly improved. This allows a HSDPA user at the cell border to enjoy a data rate close to 4 Mbps in dispersive channels. For end-user performance of TCP/IP applications, users with a dual-branch G- Rake receiver experience significantly higher packet bit rates and shorter download times.

In orthogonal frequency division multiplexing (OFDM) systems, a cyclic prefix (CP) is often used. With traditional reception, the CP is discarded. However, if the length of the CP exceeds the delay spread of the channel, a portion of the CP can be used to improve performance. In the past, Nyquist windowing techniques have been used to recover signal energy from the CP while maintaining orthogonality between symbols. In this paper, a maximum-likelihood (ML) approach is developed, which improves performance further, particularly when the signal-to-noise ratio (SNR) is low. An approximate ML solution is also given which employs a non-Nyquist window.

We develop a parametric receiver for packet data systems based on the generalized Rake (G-Rake) formulation. The focus here is on multi-antenna receivers to exploit both the spatial and temporal dimensions in suppressing interference. We extend previous G-Rake formulations to multi-antenna reception and propose an adaptive, parametric form to handle the (potentially) rapid interference variation in packet data systems. This parametric form estimates an impairment covariance matrix using a model, and calculates the model parameters based solely on short-term received data. The parameter calculation involves least-squares fitting of a short-term estimated impairment covariance matrix to the model. Simulation of a two-antenna G-Rake receiver shows that significant gains in performance are possible in both the uplink and downlink

In the downlink of direct-sequence code-division multiple-access (DS-CDMA) cellular systems, system capacity is limited by multiuser interference. This interference can be suppressed at the receiver by interference whitening approaches such as chip equalization and generalized RAKE (G-RAKE) reception. In this paper, we show how these advanced receivers can increase system capacity. First, the G-RAKE receiver formulation is extended to include soft handoff and other-cell interference. Then, voice capacity is evaluated, including the effects of fast fading, interference correlation, power control, orthogonality properties of own-cell signals, and soft handoff. Numerical results for the wideband CDMA system show that when the channel is dispersive, advanced receivers improve capacity by 30-34%

In direct-sequence code-division multiple-access (DS-CDMA) systems, signal quality estimation is an important receiver function used for power control and rate adaptation. Signal quality is usually measured in terms of a signal-to-interference ratio (SIR). Typically, a short term signal power estimate and a long term interference power estimate are used. As a result, estimation error is mostly due to the short term signal power estimate. In this paper, we consider both subtractive and multiplicative forms of bias removal for SIR estimation, and show that multiplicative bias removal provides lower root-mean-square (rms) error. These bias removal forms are developed for a traditional SIR estimation approach that has been extended to account for interference correlation across Rake fingers due to receive filtering or dispersion. The extension provides an SIR estimation approach suitable for use in conjunction with generalized Rake reception, which also accounts for interference correlation.

This paper deals with finder placement for Rake receivers, which is a key aspect in the design of receivers for spread-spectrum communications. Finger placement is typically performed using a path searcher, which estimates the power/delay profile (PDP) of the received signal to determine where signal energy is present. A commonly used finger placement approach is to place fingers at delays corresponding to the peaks of the PDP, with a restriction that fingers he no closer than some minimum spacing. Another approach is to place fingers on an equally-spaced grid of delays. For the typical urban (TU) channel model, we quantify performance of the two approaches in terms of acquired signal-to-noise ratio (SNR) and modem bit- error-rate (BER). We show that performance depends primarily on minimum finger spacing and number of fingers. The grid approach performs well when there are enough fingers to collect most of the signal energy. For a channel model in which path delays vary dynamically, we find that the grid approach provides robust performance.

In mobile communication systems, downlink (forward link) system capacity is limited by the ability of mobile receivers to recover the desired signal in the presence of cochannel interference (CCI). Joint detection of the desired and cochannel signals is a useful approach to improving receiver performance, thus increasing system capacity. In this paper, we show that a practical single-antenna joint-detection receiver can provide significant gains in system capacity for the time-division multiple-access (TDMA) standard Telecommunications Industry Association/Electronic Industry Association/Interim Standard-136 (TIA/EIA/IS-136 or IS-136). For a sectorized system, joint detection provides a capacity gain of 47% in a typical urban environment. When used in conjunction with transmit beamforming, the synergy between the two approaches leads to a capacity gain of over 200%. In determining these gains, practical aspects of the IS-136 system are considered, namely, unsynchronized networks, limited receiver complexity, and adaptability. A semiblind acquisition process, which uses the training sequence of the desired user only, is employed, because the desired and interfering base stations are not synchronized. The receiver complexity is controlled by processing only one sample per symbol period, even though it is shown that multiple samples per symbol period should ideally be used. Finally, because receiver performance may be limited by its own intersymbol interference instead of CCI, an adaptive joint-detection process is used which selects between joint demodulation and single-user equalization for each slot.

Interference suppression at the receiver can be used to improve performance and capacity in the downlink of direct-sequence code-division multiple-access systems. In this paper, an interference suppression technique is developed which uses partial knowledge of spreading sequences to cancel a group of interfering signals. Specifically, knowledge of a complex scrambling sequence is used to project the desired signal away from the interference in the in-phase/quadrature (I/Q) complex plane. A maximum likelihood receiver formulation is used, treating both the desired signal and interference as being conditionally noncircular. A zero-forcing equalization approach is used to preserve orthogonality between signals from the same base station. The development of the technique is based on the assumption that the group of signals corresponds to another base station that does not transmit the signal of interest. However, the technique is also applied to the path diversity scenario (dispersion, soft handoff, transmit delay diversity), in which the group of interfering signals includes a desired signal component. The approach, referred to as I/Q projection, provides significant gains when performance is interference limited.

The benefit of generalized RAKE reception to downlink system capacity is studied. System capacity is evaluated for certain cell and user outage probabilities. Fast fading, noise correlation, power control, and soft handoff are all included in our study. The site selection diversity transmission (SSDT) approach to power allocation between soft handoff links is shown to be optimal in the sense that it minimizes total transmit power to a given user. Numerical results show that G-RAKE receivers can improve system capacity by as much as 30-45% in a highly dispersive channel.

Abstract Channel estimation is an integral part of standard adaptive receiver designs used in narrowband, digital wireless communication systems. In this tutorial paper, commonly used approaches to channel estimation are reviewed. Both time‐invariant and time‐varying channels are considered. For time‐varying channels, both pilot symbol interpolation and data‐directed channel tracking are considered. Applications include the Global System for Mobile communications, the Enhanced Data rates for Global Evolution system, and another Time‐Division Multiple‐Access system known as Telecommunications Industry Association/Electronics Industry Association/Interim Standard—136 (TIA/EIA/IS‐136 or IS‐136). Copyright © 2001 John Wiley & Sons, Ltd.

Currently, a global third-generation cellular system based on code-division multiple-access (CDMA) is being developed with a wider bandwidth than existing second-generation systems. The wider bandwidth provides increased multipath resolution in a time-dispersive channel, leading to higher frequency-selectivity. A generalized RAKE receiver for interference suppression and multipath mitigation is proposed. The receiver exploits the fact that time dispersion significantly distorts the interference spectrum from each base station in the downlink of a wideband CDMA system. Compared to the conventional RAKE receiver, this generalized RAKE receiver may have more fingers and different combining weights. The weights are derived from a maximum likelihood formulation, modeling the intracell interference as colored Gaussian noise. This low-complexity detector is especially useful for systems with orthogonal downlink spreading codes, as orthogonality between own cell signals cannot be maintained in a frequency-selective channel. The performance of the proposed receiver is quantified via analysis and simulation for different dispersive channels, including Rayleigh fading channels. Gains on the order of 1-3.5 dB are achieved, depending on the dispersiveness of the channel, with only a modest increase in the number of fingers. For a wideband CDMA (WCDMA) system and a realistic mobile radio channel, this translates to capacity gains of the order of 100%.

Both IS-136 and PDC digital cellular systems employ forward error correction (FEC) encoding followed by a form of DQPSK modulation. To optimize FEC decoder performance at the receiver, soft information from the DQPSK demodulator is required. For frequency nonselective fading channels, which occur often in these systems, coherent demodulation optimizes performance. In this paper, we derive the maximum a posteriori (MAP) bit estimate for DQPSK modulation in non-ISI channels, assuming knowledge of the channel response. The MAP estimator forms a bit log-likelihood ratio, which provides the optimal "soft information" for MLSE or MAP convolutional decoding. Both single and multiantenna receivers are considered. MAP estimation requires exponentiation and logarithm operations, as well as knowledge of the noise covariance. To reduce complexity, two approximate forms are developed which avoid the exponentiation and logarithm operations. If the noise has the same power on each antenna and is uncorrelated between antennas, then knowledge of the noise covariance is also not needed. Performance of the optimal and approximate schemes is evaluated via simulation for single and multi-channel receivers under static and frequency non-selective Rayleigh fading conditions. The approximate approaches are shown to be within 0.5 dB of the optimal approach and to outperform existing approaches.

The capacity and signal quality of wireless communication systems can be improved dramatically by enhancing the receiver performance in the presence of adjacent channel interference (ACI). The sequence error rate is minimized through joint demodulation of all signals, but with high implementation complexity. Complexity can be greatly reduced through the use of successive cancellation, in which signals are detected and subtracted in descending order of signal power. However, successive cancellation performs poorly when signals have similar power levels. In this paper, an improved successive cancellation receiver for adjacent channel signals is proposed. The conventional detector is replaced with a simple form of joint demodulation. As a result, the receiver combines the benefits of joint demodulation, which works well when signals have similar power levels, with the benefits of successive cancellation, which works well when signals have dissimilar power levels. Performance results for the Global System for Mobile (GSM) communication digital cellular system show an order of magnitude reduction in bit-error-rate when adjacent channel signals are closely spaced and have comparable power levels.

In practical direct sequence spread spectrum (DS-SS) receivers a frequency offset between the received carrier frequency and the receiver oscillator often exists. This offset is corrected after the DS-SS signal is acquired. This paper considers the effect of frequency offset on the acquisition performance of DS-SS systems in frequency non-selective fading channels. A rapid acquisition system employing sliding correlation (matched filter) is utilized. To combat the effect of frequency offset, the total length of the sliding correlator is divided into shorter sections. We derive analytical expression for the probability of detection when the fading process is constant across all the sections. This is the case in most practical channels. The squared magnitude of the correlation results of each section are added (which we denote as non-coherent combining). The effect of frequency offset on the the probability of detection, probability of false alarm and the mean acquisition time are analyzed, allowing the trade-off between coherent and non-coherent correlation to me optimized.

The capacity and signal duality of wireless communications systems can be improved by enhancing receiver performance in the presence of adjacent channel interference (ACI). In this paper, an MLSE receiver for joint demodulation of adjacent channel signals is developed. This receiver provides significant improvement in performance over a wide range of carrier to interference ratios. A practical approach is taken, in which conventional down-conversion of each signal is performed followed by joint demodulation at baseband. To provide coherent reception, a novel form of joint channel estimation is used, exploiting knowledge of the transmit and receive filter responses. Receiver performance is evaluated via simulation for the GSM digital cellular system.

Array processing is a promising approach for improving quality, coverage, and capacity in digital cellular communication systems. By combining array processing with maximum likelihood sequence estimation (MLSE), intersymbol interference (ISI) introduced by multipath propagation can be mitigated as well. Novel symbol-spaced and fractionally spaced adaptive array processing MLSE receivers are developed for both diversity and phased array antenna configurations. The practical issues of synchronization and channel estimation are addressed. A novel approach to automatic frequency error correction (AFC) is proposed and is shown to be critical when cancelling cochannel interference. Performance is evaluated for the reverse link of the IS-136 TDMA-based digital cellular system. Substantial improvements are obtained over conventional antenna configurations for receiver sensitivity (2.5-4 dB) and over traditional antenna combining when cochannel interference is present (0.5-25 dB).

Forney (1972) and Ungerboeck (1974) have each developed maximum-likelihood sequence estimation (MLSE) receivers for intersymbol interference (ISI) channels. The Forney receiver uses a whitened matched filter, followed by a sequence estimation algorithm using the Euclidean distance metric. The Ungerboeck receiver uses a matched filter, followed by a sequence estimation algorithm using a modified metric. A unified development of both receivers is given, in which each receiver is derived from the other. By deriving the Ungerboeck receiver from the Forney receiver, we show that the whitening operation is cancelled in the Euclidean distance metric, leaving the modified metric. In addition, the Ungerboeck receiver is extended to the case of a time-varying known channel. When the channel is unknown, decision-directed channel estimation is assumed, which requires channel prediction to account for the decision delay. It is shown that the Ungerboeck receiver requires additional channel prediction, degrading performance due to prediction uncertainty. To solve this problem, two alternative receiver forms are developed which do not require additional prediction, though the computational complexity is increased. Performance and complexity of the receiver forms are compared for the IS-136 digital cellular time-division multiple-access (TDMA) standard.

The goal of many cellular system studies is the capacity evaluation and comparison of planned cellular systems. The capacity definition includes multiple quality requirements which must be met, such as a blocking probability for new calls and an outage probability for connected calls. The reuse factor and Erlang loading of each cell are adjusted so that all quality criteria are met. However, some quality criteria are exceeded. This makes capacity comparisons difficult, since systems may be compared with different qualities. One of the problems is that the site reuse factor can only take on a restricted number of values, e.g. 1,3,4,7. This paper proposes a practical method for obtaining intermediate reuse values, so that multiple quality criteria can be more closely met. This is achieved by partitioning the total channel set into two subsets, each subset being allocated with a different reuse factor. Such an approach is physically realizable and straightforward to use in capacity evaluation simulations. Also, the approach can be extended to provide a continuum of reuse values. As an example, this method is applied to evaluate the capacity benefit of receive antenna diversity for the American Digital Cellular (ADC) standard.>

The authors identify several specific sources of instability in the conventional recursive least squares (CRLS) algorithm. They explore why the algorithm is numerically unstable under fixed-point precision and examine how it can be stabilized. A form of the CRLS algorithm that maintains symmetry in the inverse autocorrelation matrix estimate is first presented, and divergence phenomena are discussed. Several sources of instability are identified as well as practical techniques for combating each instability source. Experimental results that demonstrate the usefulness of these techniques are presented. Initial simulation results indicate that by compensating for the source of numerical instability, it is possible to obtain numerical stability with negligible loss in performance.>

To reduce the amount of information presented to a human observer, ORing has been used to combine frequency-time plots from several acoustic channels. ORing introduces a loss which is measured as a rise in the minimum detectable signal-to-noise ratio. Equations for ORing loss are derived for a general signal processing system with arbitrary signal and noise distributions. Results are provided for the case of Gaussian noise and Gaussian signal. Simulation is used to verify the analytical results, and to provide insight into how cross-correlated noise and signal present on multiple channels reduce ORing loss. Results suggest: 1) if signal is present on only one channel, noise correlation reduces ORing loss, but not as much as currently estimated, and 2) if noise is independent from channel to channel, signal on more than one channel also reduces ORing loss. However, the two effects are not independent. Consequently, they must be considered together when estimating ORing loss.