Abstract-For communication over doubly dispersive channels, we consider the design of multicarrier modulation (MCM) schemes based on time-frequency shifts of prototype pulses. We consider the case where the receiver knows the channel state and the transmitter knows the channel statistics (e.g., delay spread and Doppler spread) but not the channel state. Previous work has examined MCM pulses designed for suppression of inter-symbol/inter-carrier interference (ISI/ICI) subject to orthogonal or biorthogonal constraints. In doubly dispersive channels, however, complete suppression of ISI/ICI is impossible, and the ISI/ICI pattern generated by these (bi)orthogonal schemes can be difficult to equalize, especially when operating at high bandwidth efficiency. We propose a different approach to MCM pulse design, whereby a limited expanse of ISI/ICI is tolerated in modulation/demodulation and treated near-optimally by a downstream equalizer. Specifically, we propose MCM pulse designs that maximize a signal-to-interference-plus-noise ratio (SINR) which suppresses ISI/ICI outside a target pattern. In addition, we propose two low-complexity turbo equalizers, based on minimum mean-squared error and maximum likelihood criteria, respectively, that leverage the structure of the target ISI/ICI pattern. The resulting system exhibits an excellent combination of low complexity, low bit-error rate, and high spectral efficiency.

Abstract-There is a growing demand for higher data rate systems that can function in a highly mobile environment. This mandates designs suited to doubly selective channels. This paper presents a pulse-shaped frequency-division multiplexing (PS-FDM) scheme for transmission over doubly-dispersive channels. The pulse shapes are designed to yield an inter-symbol interference (ISI)/ inter-carrier interference (ICI) profile matching a given target response. The receiver relies on a highperformance/low-complexity equalizer based on the maximum likelihood (ML) criterion to reliably extract the transmitted symbols from the observations in the presence of controlled amounts of interference in the target response. In order to protect the transmitted information against sub-carrier nulls, a convolutional code is used at the transmitter. The equalizer exchanges soft information with a maximum a-posteriori probability (MAP) optimal decoder in a turbo-like fashion at the receiver. Simulations suggest that turbo-equalization with the linear complexity iterative equalizer offers significant performance enhancements over standard techniques.

No abstract

Abstract-Multi-carrier modulation (MCM) is widely studied as a communication technique for time-and frequency selective (hence, doubly selective) channels. MCM schemes are usually designed either to reduce decoding complexity, to exploit diversity gains, or to enhance spectral efficiency. However, no known scheme accounts for all three concerns. In this paper, we propose a design metric based on complexity-constrained achievable rate that accounts for all three of these concerns. We then use this metric to characterize a trade-off between achievable rate and implementation complexity, assuming frequency-domain processing at the receiver. Finally, we use this metric to compare several MCM schemes under the same level of receiver complexity.

Beamforming and combining strategies are extensively used to harness spatial diversity gains in MIMO-OFDM systems. However, traditional approaches are ineffective when time-variations in the underlying wireless channel introduce inter-carrier interference (ICI). We propose novel beamforming and combining strategies for such time-and frequency-selective wireless channels. Results show that our schemes enjoy large gains over traditional approaches. Additionally, we find that our schemes are robust to the use of predicted (as opposed to perfect) channel state information at the transmitter. 1