Optimization of the transmission range in terms of maximizing information efficiency is studied in this paper for mobile ad hoc networks (MANETs) with frequency-hopped (FH) CDMA and multiple antennas. Realistic channel models are employed to account for path-loss, log-normal shadowing, and Rayleigh fading. The shadowing and fading are assumed time-varying with different time-scales due to mobility and hopping. The receiver employs decision-feedback demodulation for differential unitary space-time modulation and erasure insertion decoding for Reed-Solomon codes [14] [15]. The decoding error probability is derived based on distributions of the multiple access interference power and the signal-to-interference ratio in a dwell for ground propagation model. The trade-off between information efficiency and transmission range is studied in detail and insight is obtained into the impact of various factors, including the spreading gain, modulation and coding schemes, feedback length, erasure insertion, order of spatial diversity, time-variation of the shadowing, and channel statistics such as Doppler frequency and shadowing spread.

In this paper, we develop a coded MIMO FH-CDMA transceiver that is robust to partial-band jamming and the near-far problem for ad hoc networks with high mobility. Spatial and frequency diversity are achieved respectively by differential unitary space-time modulation (DUSTM) and coding/interleaving. Decision-feedback demodulation (DFD) and erasure insertion (EI) decoding are employed to mitigate fast Rayleigh fading and interference. A decision-directed adaptive estimator is also developed to estimate the required parameters in DFD and EI. The two proposed EI schemes are based on the Bayesian framework proposed by Baum and Pursley (1992) and a novel threshold test of the effective signal to noise and interference ratio (ESINR), respectively. Decoding error probability of the ESINR based EI is derived analytically. Performance of both EI schemes are examined under various interference conditions, and the effect of EI thresholds is carefully discussed. The Bayesian EI is shown to be insensitive to suboptimal threshold and achieves significantly better performance than error-only decoding. Furthermore, it is shown that analysis of the ESINR based EI scheme predicts the performance of the Bayesian EI when the interference is dominant.

In this paper, we consider coded frequency-hopping spread spectrum (FHSS) systems with differential unitary space-time modulation (DUSTM). We investigate the erasure insertion (EI) techniques for improving the decoding error probability, especially in presence of strong partial-band interference (PBI). Conditional likelihood functions are derived for the decision-feedback demodulation and used for EI. The two proposed erasure rules do not require knowledge of the distribution of the noise plus interference power, in contrast to their counterparts in [14]. Instead of assuming the unknown parameters involved in decision-feedback demodulation and EI to be known, a joint estimation-demodulation scheme based on the adaptive recursive least square (RLS) algorithm is proposed. The sensitivity of the proposed EI rules to mismatched thresholds is studied by simulation. Numerical results show that EI can improve the decoding error probability significantly in presence of PBI, even when the required parameters have to be estimated and suboptimal thresholds are used.

Studies on differential unitary space-time modulation are mostly focused on Rayleigh fading channels. In this paper we present new analytic results for the maximum-likelihood demodulator (MLD) and the standard differential demodulator (SDD). The resulting expressions and bounds on the pairwise error probability shed light on the impact of various system parameters and are helpful towards a deeper understanding of the DUSTM in Rician channels. Additional analytic results are also derived for the constellation in the work of Tarokh and Jafarkhani (2000). Unlike in Rayleigh fading, the exact MLD can only be implemented approximately (Cui and Tellambura, 2006). The performance loss of practical demodulator such as the SDD and a prediction-based demodulator relative to the MLD are studied under different Rician fading environments.

We investigate two erasure insertion schemes for a coded slow frequency-hopping system with multiple antennas at both the transmitter and the receiver. The demodulated symbols may become unreliable due to fading, thermal noise, and partial band interference (PBI). Since error correcting codes can correct more erasures than errors, it is often beneficial to erase these unreliable symbols rather than feeding them to the decoder. In this paper, we first extend the Bayesian erasure insertion, originally proposed in C.W. Baum and M.B. Pursley (1992) for FSK modulation, to the case when differential space-time modulation is used. We then propose a simple, suboptimal erasure insertion scheme based on a likelihood ratio threshold test (LRTT). Simulation results show that substantial improvements in decoding error probability can be achieved by the proposed schemes over the non-erasure-insertion scheme, especially when the partial-band interference becomes severe.

We consider the problem of interference suppression in a multipath DS-CDMA downlink using a general RAKE receiver structure with M fingers and arbitrary finger delays. Utilizing asymptotically optimal sampling design theory, a new design approach that optimizes the receiver performance over the finger delays is proposed. It provides a set of delays which are asymptotically optimal for large M, and sheds light on the behavior of various finger placements' performance, as the number of fingers increases, and on the effect of the smoothness of the underlying chip pulse. Numerical results show that for certain Rayleigh fading channels and several chip pulses, the proposed finger placement is superior to the chip periodically-spaced finger placement.