Multiplex communications – Generalized orthogonal or special mathematical techniques – Fourier transform
Reexamination Certificate
1998-05-22
2002-04-02
Ton, Dang (Department: 2661)
Multiplex communications
Generalized orthogonal or special mathematical techniques
Fourier transform
Reexamination Certificate
active
06366555
ABSTRACT:
TECHNICAL FIELD
The invention relates generally to digital information transmission systems and specifically to techniques for controlling signal clipping over a discrete multi-tone communications channel.
BACKGROUND OF THE INVENTION
Discrete Multi-Tone (DMT) systems partition the available transmission bandwidth into many narrowband subchannels over which parallel data streams are transmitted. For example, in one such communications system, a DMT symbol is created by summing several signals that are modulated (e.g. Quadrature Amplitude Modulation (QAM) to different frequencies called the tones of the signal. Summing many random signals leads to a transmitted signal whose probability density function is close to Gaussian and has a much higher Peak to Average power Ratio (PAR) than most single-carrier modulated signals.
The high PAR commonly employed in a DMT system can cause clipping of the transmission signal and loss of data. A Digital to Analog Conversion (DAC) circuit can be chosen to provide significantly high resolution in an attempt to limit the effects of clipping as well as distortion at the Analog Front End (AFE) which forms the interface to the network line. The AFE must have a larger dynamic range than the signal spectrum from the output of the DAC. This is a significant disadvantage for DMT when compared to single-carrier modulation because both the DAC and AFE can become a significant percentage of the cost of the transmitter/receiver. In addition, the AFE can become the greatest power consuming element of the system.
To reduce AFE complexity and power consumption, DMT transmitters are often designed to support lower PAR values. However, this savings in complexity often causes clipping of the DMT signal, which can reduce performance. A DMT signal has a much higher peak to average ratio than a comparable single tone modulation signal. Therefore a DMT transmitter has to either have a significantly more expensive AFE both in terms of the analog filter and line drivers to effectively control the amount of clipping.
Prior methods of reducing the complexity and costs of the AFE and DAC in a DMT system include modifying the transmitted signal to reduce the amount of clipping. For example, a prior method uses signal processing in the transmitter to reduce the effect of clipping at the front end of the transmission while the decoding operations in the receiver remain unchanged. Another prior method applies signal processing algorithms in the transmitter, but depends on the receiver to recognize that clipping control has been applied to perform appropriate inverse operations.
An example of these prior methods is the spectral shaping technique described in “Mitigating Clipping Noise in Multicarrier Systems”, by J. S. Chow, J. A. C. Bingham and M. S. Flowers, 1997 IEEE International Conference on Communications. When the transmitter detects a clip it adds more “noise” to the signal to modify the spectrum of the clip and push more energy into the higher frequencies. The clip, in turn, appears as an impulse which is white noise in that part of the spectrum. In theory, as the SNR is dropping towards higher frequencies, the spectrally shaped clipping noise will have little effect on the total SNR of the signal. This method, however, has the disadvantages that the total clipping noise is increased and no attempt is made in the receiver to cancel this noise. Thus, the receiver decodes the signal in the same way whether the shaped clipping noise is present or not.
Other prior methods of dealing with signal clipping decrease the size of the input data block so it is smaller than the DMT block size. The transmitter performs a one-to-one map of the data to a subset of possible DMT blocks that are known to have good PAR values. The complexity of the mapping routines, however, makes such a technique unfeasible for many applications except for those using very small block sizes (e.g. 4 tones) since there is no straight forward way to achieve mapping in more complicated systems. Additionally, such mapping techniques often result in a significant reduction of data rate.
Another prior method of dealing with clipping includes the setting aside of tones for the transmission of information describing the actions of the transmitter. An example of such a technique is also described in the article “Mitigating Clipping Noise in Multicarrier Systems” wherein the transmitted signal is scaled down by a factor and if clipping occurs a reserved tone is used to communicate the scaling factor to the receiver. Incorrect decoding of the reserved tone, however, will lead to incorrect decoding of the whole block, so the reserved tone is a high SNR tone that is carrying only a few bits of information.
Still another method of dealing with clipping problems is disclosed in the article “A Method to Reduce the Probability of Clipping in DMT-Based Transceivers” by D. J. G. Mestdagh, P. M. P. Spruyt, IEEE Trans. On Communications, Vol. 44, No. 10, Oct. 1996, pp 1234-1238, which describes a method wherein a known random phase sequence is added to the phase of the transmitted signal if clipping occurs. For this to work, however, one must assume the probability of both the original signal being clipped and the original signal plus random phase being clipped is much less than the probability for the original signal alone being clipped. Again, a high SNR tone is used to signal that a random signal has been added.
With the development of high speed digital subscriber lines and components supporting such lines, a premium is placed on effective and reliable data high rate throughput. For example, modems supporting DMT transmissions must be able to work within the constraints of the transmission standard. With asymmetric digital subscriber line (ADSL) signaling, the modem or transceiver must be able to support a large amount of digital and analog signal processing to achieve the data throughput rates called for by the ADSL standard. Until the present invention, however, the problem of signal clipping and the desire to limit the expense in the modem made widespread deployment unfeasible especially for the home or small business user. As such, a method of controlling clipping in the transmitted signal without using a sophisticated AFE or DAC required by prior techniques is needed.
SUMMARY OF THE INVENTION
Accordingly, the present invention relates to techniques for clipping control ideal for use on any discrete multi-tone (DMT) transmission system such as asymmetric digital subscriber line (ADSL) or Orthogonal Frequency Division Multiplexing System (OFDM) such as for terrestriol broadcast in wireless transmission.
In one embodiment, a method of controlling the transmission amplitude of signals in a DMT communications system limited by a predefined dynamic range is disclosed. The method comprises the steps of converting a block of bits into a set of M constellation points in the fourier domain; and transforming the M constellation points into N complex points, wherein M is less than N, each of the N complex points falling within the transmission subspace within said predefined dynamic range.
The method can also include a mapping function that associates each M value to an N value whose real inverse fourier transform is within the predefined dynamic range. In one embodiment, the mapping function is an iterative process that starts at an initial guess of a point within the N dimensional array defined by the N complex points and converges to a solution who real inverse fourier transform is within said predefined dynamic range.
In another embodiment, the method further comprises the step of transforming the M constellation points into N complex points by the steps of: performing a one-to-one mapping of the M constellation points to the N complex points; adding an initial estimate of a displacement vector (B) to the N complex points to obtain the sum N+B; performing a real inverse fourier transform of the sum N+B; and testing the real inverse fourier transform of the sum N+B to determine if it within said predefined
Gatherer Alan
Polley Michael Oliver
Brady III W. James
Moore J. Dennis
Telecky , Jr. Frederick J.
Texas Instruments Incorporated
Ton Dang
LandOfFree
Method and device for controlling signal clipping in a... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and device for controlling signal clipping in a..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and device for controlling signal clipping in a... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2852707