| Multicarrier modulation possesses several properties which make it an attractive approach for high speed copper wire communications networks. Among these properties are the ability to e ciently access and distribute multiplexed data streams, and a reduced susceptibility to impulsive, as well as to narrowband channel disturbances. In digital implementations of multicarrier modulation, subcarrier generation and data modulation are accomplished digitally using orthogonal transformations of data blocks. These implementations are particularly e cient with regard to bandwidth utilization and transceiver complexity. In this paper, we present a form of digital multicarrier modulation which we refer to as overlapped discrete multitone, or discrete wavelet multitone (DWMT), modulation. For DWMT modulation, which is based on the application of M-band wavelet lters, the pulses for di erent data blocks overlap in time, and are designed to achieve a combination of subchannel spectral containment and bandwidth e ciency that is fundamentally better than with other forms of multicarrier modulation. We show that, as a result of the spectral containment feature, DWMT gives a high level of robustness with regard to noise environments and channel variations that are encountered in practice.

No abstract

A multicarrier niodulation system called Discrete Multitone (DMT) was recently chosen by the American National Standards Institute ( A N I ) as a working standard for asymmetric digital subscriber line (ADSL) signaling. The DMT system use5 the inverse and forward discrete Fourier transform (DFT) for modulation and demodulation. An alternative to the classic DAlT system is to substitute a discrete wavelet transform in place of the DFT as the modulating transform. This new method is called Discrete Wavelet Multitone (DWMT) signaling. and has been shown to be superior tao the D31T system in certain channel environments. This paper illustrates that the DMT system and the DlYILIT system are both based on the theory of AI-band wavelet filter baaks. In particular. DhIT and D\l'?rlT are linked formall) through a discussion of digital niultirate filter banks and digit a1 transniultiplexers. In addition the advantages of wavelet transform-bawd t ransmultiplesers over DFThased transniult iplesers are discussed.

Conventional detection and demodulation rules are not ideally suited for use with discrete Fourier transform (DFT) based narrow-band interference suppression, if the DFT incorporates time-weighting to localize the interference spectrally. Detection rules which are adapted to the time-weighting are introduced here and shown to o er signi cant improvements in performance, but their implementation is complex. It is found that if a spectrally-contained orthogonal transform (SCOT) is employed in place of the windowed DFT however, the conventional detection rules have a level of performance comparable to that obtained with DFT based suppression and the adapted rules. In addition to yielding good interference localization, SCOTs bene t from the fact that they provide a Karhunen-Loeve like transform for each component of the observation. The primary focus of the paper is the application of interference suppression for the detection of the presence of covert signals; however, it is also demonstrated that SCOT based suppression is potentially better suited than DFT based suppression for data demodulation in commercial code-division multipleaccess (CDMA) network overlays.

No abstract