Wavelet Applications in Communications

OFDM Review

OFDM or Orthogonal Frequency Division Multiplexing is a data transmission scheme that consists of transmitting a high rate data stream by using lower rate data streams modulated over orthogonal subcarriers. An OFDM transmission system consists of a constellation encoder followed by a serial-to-parallel converter and a FFT encoder which modulates the signal symbols over an orthogonal bank of subcarriers. The stream is then converted from parallel to serial and a redundant guard band is added to prevent Inter Symbol Interference caused by multipath propagation delays in wireless and other channels.Block diagrams of OFDM Transmitters and Receivers are shown in figures below.


OFDM Transmitter


OFDM Receiver

OFDM Stages

On the transmitter side, an OFDM system consists of

OFDM receiver stages are essentially the reverse of the transmission stages as can be seen from the block diagram. Mathematically,an OFDM signal \(x[k]\) can be seen as a series of ,say, \(s\) OFDM symbols each consisting of \(M\) orthogonally modulated waveforms.

\[ x[k]=\sum_{s}\sum_{m=0}^{M-1} a_{s,m} \phi_{m}[k-sM] \]

where \(\phi_{m}[k]\) are \(M\) orthogonal waveforms such that

\[ \lt\phi_{m}[k],\phi_{n}[k] \gt = \delta[m-n] \]

In the case of regular OFDM case, we use FFTs to implement orthogonal channels, such that

\[ \phi_{m}[k]=e^{\frac{j2\pi mkT}{M}} \]

Advantages of OFDM

Wavelet Packet Modulation

Since OFDM channels consist of \(M\) orthogonal subcarriers, we need \(M\)-band wavelet implementation if we want to substitute wavelet transform in place of DFTs. We achieve this using wavelet packet decomposition and reconstruction. Instead of dyadic iterations across only low pass filters, we dyadically decompose both high and low pass filters at every stage as show in the figure for two level wavelet packet tree decomposition.


Wavelet Two-level Packet Decomposition

This wavelet tree is equivalent to one stage \(4\) band wavelet filter bank as shown in the figure.


WaveletTwo-level Packet Decomposition Equivalent Representation

Decomposition is done similarly for any \(M=2^{n}\). The filters of \(M\)-band wavelet transform are obtained using noble identities and iterated filter bank properties. They are respectively \(H(z)H(z^{2})\),\(H(z)G(z^{2})\),\(G(z)H(z^{2})\) and \(G(z)G(z^{2})\). Wavelets corresponding to these filters are computed using scaling and wavelet dilation equations. The dilation equations are different in the \(M\)-band case:

\[ \phi(t):= \sum_{k} hh(k)\phi(Mt-k) \]

and

\[ \psi(t):= \sum_{k} gg(k)\phi(Mt-k) \]

where \(hh\) and \(gg\) filters are iterated bandpass filters obtained using noble identities.

An \(M\)-band wavelet system has one scaling function corresponding to the low pass filter and \(M-1\) wavelet functions corresponding to each bandpass filter. For a three level wavelet packet decomposition using Daubechies orthonormal wavelets, the filters and \(M = 8\) wavelets[unnormalized] are plotted in Matlab.


8 Filters for 3-level Packet Decomposition Tree


Wavelets for 3-level Packet decomposition Tree

Wavelet Packet Based OFDM systems

Wavelet based OFDM systems are similar to regular OFDM with orthogonal subcarriers generated by iterated wavelet filter banks instead of DFTs but there are several crucial differences. Wavelet-OFDMs have better ISI and ICI performance. In addition, wavelet symbols generated are longer than DFT symbols. The length of iterated wavelet filters for \(M\) level wavelet packet decomposition is equal to \((M-1)(L-1)+1\) where \(L\) is the length of original filters. For higher level decompositions, iterated wavelet filters are significantly larger which results in longer subcarrier symbols. Guard Band or cyclical prefix is usually not needed in Wavelet OFDM case. A block diagram of Wavelet OFDM can, therefore, be given by


Wavelet-OFDM Transmitter


Wavelet-OFDM Receiver

Some properties of W-OFDM are following

For more information on WOFDM applications in industry, most prominently in wireless and powerline communications, see HD-PLC Panasonic, IEEE 802.11 standard resources( for information about conventional OFDM) and IEEE P1901 powerline standards/working groups websites. Check the references.

Navigation

Fourier Transform

Discrete Wavelet Transform

Filter Banks and Wavelets

Iterated Filter Banks

Resources

IEEE 802.11:WLAN

IEEE P1901:Powerline Group

HD-PLC Alliance

USF OFDM and MC-CDMA Page

WPT Java Implementation

Wavelet Paket Transformation [German]

References

Mallat::A Wavelet Tour of Signal processing

IEEE 802.11a-1999 Standard

RM Wavelet Based (WOFDM) PHY Proposal for 802.16.3, Rev.0.0

Jamin, Mahonen::Wavelet Packet Modulation for Wireless Communications

A.L.Intini :: Orthogonal Frequency Division Multiplexing for Wireless Networks

Vaidyanathan::Multirate Systems and Filter Banks

Fisal, Hosseini ,Syed-Yosef::Wavelet Packet based Multicarrier Modulation for Cognitive UWB Systems

Galli, Koga, Kodama :: Advanced signal processing for PLCs: Wavelet-OFDM

Weimin Yang, Guangguo Bi, Yum, T.-S.P.::A Multirate Wireless Transmission System Using Wavelet Packet Modulation