Pulse or digital communications – Receivers – Interference or noise reduction
Reexamination Certificate
2001-08-21
2004-04-20
Bayard, Emmanuel (Department: 2631)
Pulse or digital communications
Receivers
Interference or noise reduction
C375S229000
Reexamination Certificate
active
06724841
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to a method for processing received signals in order to remove waveform distortion, and also to a corresponding device. In particular, the present invention is directed to a method and corresponding device used for improving an equalizer in a receiver part of digital telecommunication systems.
BACKGROUND OF THE INVENTION
In the last years, telecommunication systems and especially wireless telecommunication systems using digital data transmission methods are widely spreading. The usage of such digital data transmission methods allows to transmit data at a high transfer rate. In case of e.g. a mobile telecommunication system such as GSM (Global System for Mobile communications) it is possible to transmit speech or data between a stationary transceiver unit, i.e. a base station, and a terminal device, such as a mobile station, in a circuit switched mode as well as in a packet data mode. Further developments such as HSCSD (High Speed Circuit Switched Data), GPRS (General Packet Radio Service), EDGE (Enhanced Data Rates for GSM Evolution) and the like provide more sophisticated data transmission performance.
However, on the receiving side (i.e. the mobile station or the base station), the incoming signals received for example via an antenna suffer from waveform distortion caused for example by multipath propagation. Such waveform distortion like intersymbol interference and additive noise makes it more difficult for the receiving side to assign, identify and reconstruct the received signals (i.e. the symbols) correctly, since a mixture of signals is received with different delay times and amplitudes. In particular in high speed digital data transmissions, this is problematic.
To remove waveform distortions, in telecommunication systems, the usage of equalizers at the receiving side is commonly known. In said equalizers, on the basis of e.g. a channel impulse response, a decision is made how the incoming signals are to be interpreted, i.e. how the symbols are to be detected from said mixture of received signals.
Hitherto, several solutions for different types of equalizers are known to be usable in telecommunication systems. Some of them are for example described in “Introduction To Mobile Communication” by Y. Akaiwa, pages 276 to 287, John Wiley & Sons, New York, USA, 1997, in “Delayed decision Feedback Sequence Estimation” by A. Duel Hallen and C. Heegard, IEEE Transactions on Communications, vol. 37, no. 5, May 1989, in “Reduced-State Sequence Estimation With Set Partitioning And Decision Feedback” by M. Vedat Eyuboglu and Shahit U. H. Qureshi, IEEE Transactions on Communications, vol. 36, no. 1, January 1988, in “MMSE Decision Feedback Equalizers: Finite-Length Results” by N. Al-Dahir and John M. Cioffi, IEEE Transactions on Information Theory, vol. 41, no. 4, pages 961-975, July 1995, and in “Fast Computation Of Channel Estimate Based Equalizers In Packet Data Transmission” by N. Al-Dahir and John M. Cioffi, IEEE Transactions on Information Theory, vol. 43, no. 11, pages 2462-2473, November 1995.
The main purpose of such an equalizer is to reconstruct the received signal in such manner that it is as similar as possible to the original signal. This can be achieved for example by estimating the channel impulse response and use it to reconstruct the received signal. In general, it is possible to use a known training pattern (i.e. a training sequence) included for example in each sent data packet to estimate the channel impulse response at the receiving side. Then, settings for the equalizers (i.e. of equalizer or filter taps) can be computed. For a good performance, the duration of the training sequence has to be short.
One proposed solution for an equalizer is a maximum-likelihood sequence estimation (MLSE) using a trellis-based Viterbi algorithm. A simplified structure of such an equalizer is shown in FIG.
3
A. Further to the MLSE part, a channel estimator for estimating a time-varying channel impulse response is required. In general, the MLSE equalizer shows almost optimal detection performance. However, particularly in multilevel modulation system, this method becomes unpractical for use since its complexity is proportional to the number of states in trellis which increase significantly when multilevel modulation is used. For example, in case of an 8PSK (8 level phase shift keying) modulation and in a possible environment where 5 to 6 taps may be used, the number of states is several thousands. Therefore, the complexity of such MLSE equalizers increases to such an extent that present implementations for mobile station or base station equalizers are not able to manage it, or the costs of such an equalizer would not be economical.
In order to get a less complex equalizer, though with less optimal performance, there are given several other solutions in the prior art. In one of said solutions a reduced-state sequence estimation (RSSE) is used. This method is similar to the above described MLSE and is also based on the Viterbi algorithm. However, in comparison to the MLSE, the RSSE uses trellises with a reduced number of states, which leads to a less complex operation. The principle structure is shown in FIG.
3
B. Here, a feedforward filter is additionally used by which a channel impulse response is shaped to a minimum phase.
It is also known to equalize incoming signals by using a decision feedback equalizer (DFE). In this case, only a part of the impulse response is “open for decision” in the equalizer. By feeding back results of decisions of the signals, the energy of the rest of the taps (which are “not open for decision”) of the equalizer can be reduced. In general, the effective signal energy and therefore the effective signal-to-noise ratio (SNR) is defined by the energy in those impulse response taps which are “open for decision” in the trellis. A method for maximizing this energy portion is to pre-filter the incoming signals by which the channel impulse response is shaped to a minimum phase. This will maximize the effective SNR for signal detection, i.e. for the decision.
The principle structure of such a DFE is shown in FIG.
3
C. Here, also after passing the incoming signal through a feedforward filter which shapes the channel impulse response to the minimum phase, the filtered signal is fed to a decision part in which a decision about the signal (or symbol) is made. Then, the decided output of the decision part is fed back via a feedback filter for eliminating the effect (e.g. an intersymbol interference) of previously detected (i.e. decided) symbols on the decision of the current (next) symbol or signal.
For the filters, in the DFE, as well as in the above described RSSE case, most preferably finite impulse response (FIR) filters are used. Such FIR filters exhibit good numerical properties and lend themselves to an easy adaptive implementation.
As a performance criterion in the DFE, for example, a minimum mean square error (MMSE) can be used, which results in a so-called MMSE-DFE equalizer. The MMSE criterion (or cost function) can be written as
J=E|x
i
−yf−xb|
2
, (1)
wherein J is the MMSE criterion (or the cost function result), E indicates an expectation value, x
i
is a data symbol (incoming signal) currently to be estimated, y is a vector containing the received signals or samples, f is a column vector containing feedforward filter taps, x is a data symbol vector containing symbols decided before x
i
, and b is a column vector containing feedback filter taps.
In the MMSE-DFE equalizer, J has to be minimized with respect to f and b. Now, the taps for the feedforward and feedback filters can be calculated, as for example described in the references mentioned above.
However, the above described solutions suffer from several drawbacks. As mentioned above, the MLSE leads to complex calculations or equalizer structures which make it not useful for e.g. multilevel modulation systems. Though the RSSE and MMSE-DFE solutions are easier to implement, here another problem occu
Bayard Emmanuel
Nokia Corporation
Squire Sanders & Dempsey L.L.P.
LandOfFree
Equalizer with a cost function taking into account noise energy does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Equalizer with a cost function taking into account noise energy, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Equalizer with a cost function taking into account noise energy will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3190747