We propose a simple low-complexity iterative random search (IRS) algorithm based on dithering for optimal one-bit precoding, to find transmitted signals for a base station equipped with one-bit transmitters to serve multiple users in massive MIMO downlink. The complexity increases linearly with the number of transmitter antennas and we may even achieve the global optimal performance in the worst case single user scenario. The analysis shows that the error vector magnitude (EVM) of the signals arrived at the users is inversely proportional to the square of the ratio between the number of transmitter antennas and the number of the users. We can achieve about -40 dB EVM when the ratio is 200.

We analyze the resiliency of Massive Multiple-Input Multiple-Output (M-MIMO) systems to Inter-Symbol Interference (ISI) when diversity combining techniques are used at the Base Station (BS). We show that Maximum Ratio Combining (MRC) alone can equalize an ISI channel as the number of antennas grows unbounded. Additional constraints on the nature of the channel must be postulated depending on whether the information of the Angle-of-Arrival (AoA) is exploited at the receiver or not. Interestingly, the simpler Equal Gain Combiner (EGC) receiver is also able to equalize the channel as the number of antennas grows but, in this case, at least one channel path must be Ricean faded. These findings are confirmed via simulation on WSSUS channels and channels generated with a ray tracing engine simulating a real BS deployment in downtown Hong Kong and Shanghai. Finally, the observed scaling law indicates that normalized ISI power decreases N-fold for every N-fold increase in the number of antennas at the BS.

Received Signal Strength (RSS) geolocation is investigated under a Correlated Log Normal (CLN) model. Several parameters in the model may be unknown and we study their impact on both position estimation algorithms and the Cramer-Rao bound. Nuisance parameters in the CLN model include the pathloss exponent, the shadow fading standard deviation, and the shadow fading correlation. To estimate the user position, we compare the Steepest Descent and the Levenberg-Marquardt (LM) algorithms. Objective functions with and without weighting by the cova-riance matrix are considered. We prove from the CRLB that knowledge of the fading deviation and correlation is not needed by an optimal estimator. Results show the weighted objective function is a better metric, and that the LM algorithm converges very rapidly, nearly attaining the CRLB.

Cramer-Rao bounds for geolocation based on Received Signal Strength have previously been studied under a Log Normal fading model. This paper develops new bounds under several different assumptions to identify the most promising scenarios. We find the CRLB under a spatially Correlated Log Normal (CLN) model. We derive a new bound for a Relative Received Signal Strength (RRSS) technique based on power ratios. Finally, for moving receivers, we model data collection over a time window where multiple independent fading instantiations are observed. Results show that RSS location accuracy improves significantly (25%) in realistic fading correlation, and more under stronger correlation. We prove that RRSS geolocation is no better than RSS in CLN fading. For moving users, processing a time window of 20 samples reduced location error by up to a factor of 4.

The Generalized-Rake (G-Rake) receiver consists of a group of despreaders (fingers) with MMSE combining. Several techniques for assigning the finger delay positions based on the current channel impulse response have been proposed in the literature. We compare the performance of several placement techniques in the context of the UMTS/HSPA uplink. Two new finger placement techniques are proposed: a simple "equalizerlike" placement in the time-domain, and an FFT-based approach that exploits the ease of computing the Weiner filter in the frequency-domain. Additionally, we assess two methods for finding the required covariance matrix; a traditional approach based on the sample covariance matrix, and a "formula-based" approach that uses the channel estimates and exploits the known matrix structure. While most G-Rake studies have focused on the downlink, we show that the G-Rake can achieve useful 1-2 dB gains for HSUPA uplink signals with low spreading gain on dispersive channels.

Development of future geo-location systems requires an understanding of the fundamental limits imposed by system parameters such as bandwidth, SNR, number of sensors, etc. Additionally, the system designer can choose among several existing location techniques, such as time-of-arrival (TOA), time-difference-of-arrival (TDOA), and round-trip time-of-arrival (RT-TOA). This paper presents new Cramer-Rao bounds for the RT-TOA case and provides fair numerical comparisons of the bounds for these three time-based location methodologies. Our comparisons assume independent, flat Rayleigh fading on each sensor link, different average link SNRs based on path length, and a uniform user position distribution. We find that TOA outperforms TDOA and RT-TOA, while the relative ordering of the latter two depends on the sensor geometry.

It is commonly believed that OFDM systems have a complexity advantage over CDMA. We show that there are two ways for CDMA complexity to remain competitive with OFDM: a multi-channel approach, in which a wideband system is composed of multiple narrow-band systems; and a frequency-domain equalization approach, where the CDMA receiver is based on the same efficient FFT technique used in OFDM. By making some assumptions on receiver design and wireless channel characteristics, we find simple formulas for the required number of complex multiplications per second. MIMO and broadcast systems are considered as special cases.

Wireless CDMA cellular downlink communications with transmit antenna arrays in multipath fading channels is studied. Transmit diversity and various beamforming techniques are investigated and compared, in conjunction with power control. No instantaneous downlink channel information is assumed; however, the obtained results are also compared with results assuming ideal feedback. The study is carried out for both circuit-switched and packet-switched systems, for which different conclusions are drawn.

We consider the downlink rate scheduling problem for CDMA networks with multiple users carrying packets with different types of traffic ('voice/audio only', 'data-only' or mixed traffic), each of which has its own QoS requirements that need to be satisfied. At each timeslot we serve the user with the earliest deadline. If his head packet was fully served and there is still power left in the system, we repeat the procedure with the next earliest deadline user, as long as we have power resources. We can thus serve more than one user per timeslot, power permitting. We compare our proposed algorithm to the earliest deadline one, the best scheduling scheme presented by Andrews, Kumaran, Ramanan, Stolyar and Whiting (see Bell Labs Technical Journal, 1999) for similar situations and conclude that we outperform it.

We present simulated error rate and code delay/amplitude tracking performance for two recently developed detectors: the adaptive whitened matched filter, and the quasi-synchronous decorrelator. Rapid convergence (within 50 symbols) of the estimated amplitude and delay of the desired user is demonstrated based on an approximate joint maximum likelihood approach. The adaptive whitened matched filter exhibits better bit error rate performance than the decorrelator, but requires a complex eigenanalysis stage.

Third generation CDMA systems will simultaneously support traffic with different data rates, such as data and voice traffic. Advanced space-time processing can be used to achieve high system capacity in such systems. In this paper, we extend the previously developed multiple antenna successive interference canceller of Mailaender and Huang (see UCSD Conference on Wireless Communications, p.47-53, San Diego, CA, 1999) to the case of mixed data rate operation. The gain in this case relative to a simple space-time RAKE receiver is even greater than in the single rate case.

A linear decorrelator detector is proposed for a quasisynchronous code-division multiple-access (QS-CDMA) cellular system. It is assumed that each of N users has a Global Positioning System (GPS) generated local clock and attempts to transmit synchronously with the other users in its cell. The base station thus receives N signature waveforms which are approximately synchronized, but with a small timing error due to local oscillator drift, Doppler shifts, and variable distances from the mobiles to the base. The residual timing offset is discretized, and a decorrelator is constructed which attempts to reject the multiuser interference vectors corresponding to the set of possible offsets. A noncoherent version of the decorrelator detector is presented, along with a maximum-likelihood scheme for synchronizing to the desired signal. Bounds and analytical results for the bit-error rate (BER) are obtained for various interferer conditions.

A multiuser detector is developed in which the delays and amplitudes of the incoming waveforms are estimated recursively. The algorithm is an extension of the symbol-by-symbol detector of Abend and Fritchman, originally derived for intersymbol interference (ISI) channels, to the multiuser application. In order to make the multiuser detector adaptive, the likelihoods in the symbol-by-symbol metric update are approximated using a set of extended Kalman filter (EKF) innovations. The EKF's provide, in addition to the likelihoods, joint estimates of the signal delays and amplitudes. The resulting algorithm is quite complex, due to the large number of possible composite symbols corresponding to the multiple users. However, it is shown that the likelihood computations and EKF updates can be expressed it terms of a set of cross-correlation functions, which need only be computed for a subset of the possible symbols. The cross-correlator outputs are updated at the bit rate, and thus the EKF and metric update computations need be performed only at a function of the spread-spectrum chip rate. A metric pruning technique is proposed that further reduces the number of EKF delay/amplitude estimators required. Finally, an important sampling simulation strategy is used to obtain bit-error rate estimates for the adaptive multiuser detector.>