Pulse or digital communications – Receivers – Interference or noise reduction
Reexamination Certificate
1999-03-15
2001-06-26
Chin, Stephen (Department: 2634)
Pulse or digital communications
Receivers
Interference or noise reduction
Reexamination Certificate
active
06252918
ABSTRACT:
BACKGROUND
The present invention relates generally to communication systems and, more particularly, to coherent demodulation in digital communication systems.
Digital communication systems are developing rapidly for both wireline and wireless applications. In such systems, information is converted to information symbols, typically binary in value. These symbols are then encoded and modulated to a form that can be transferred onto a transmission medium, such as wires, the air (using radio waves or acoustic pressure), or magnetic tape. This transmission process includes the use of pulse shaping, to shape the symbol values for transmission.
As the demand for communications grows, higher data rates are employed in wireline modems. As a result, the modulation and/or transmission medium introduces intersymbol interference (ISI), in which one symbol overlaps with the next. To provide acceptable performance, equalization at the receiver is required, such as linear equalization, decision feedback equalization, or maximum likelihood sequence estimation (MLSE).
Equalization is also required in many wireless communications systems. These systems include those defined by D-AMPS, PDC, and GSM, which employ a combination of frequency division multiple access (FDMA) and time division multiple access (TDMA). Therein, the transmission medium is dispersive due to multipath propagation, giving rise to ISI at the receiver.
New forms of modulation and multiple access have been introduced in wireless communications, such as spread spectrum modulation and code division multiple access (CDMA). In these types of systems, for example, those defined by IS-95, an information symbol is represented by a sequence of chips (modulation symbols). As with FDMA and TDMA, multipath propagation in CDMA systems gives rise to signal echoes at the receiver. In a CDMA system, ISI is handled with a Rake receiver.
In all of these systems, speech quality is significant to customer satisfaction. To provide high speech quality at the receiver, advanced demodulation techniques are used in the form of, for example, an equalizer or a Rake receiver. These advanced demodulation techniques require estimating channel tap coefficients, which correspond to different signal delays. These channel tap coefficients are then used in the demodulation process.
Demodulation is typically performed at baseband. In a radio receiver, this occurs after the received signal has been filtered, amplified, mixed down to the baseband frequency, sampled and quantized. This results in a stream of received samples, denoted r(k), which are traditionally modeled as:
r
(
k
)=
c
(0)
a
(
k
)+
c
(1)
a
(
k−
1)+ . . . +
n
(
k
) (1)
where c(j) are the channel tap coefficients and a(k) are the transmitted symbol values. Complex values are assumed, which correspond to in-phase (I) and quadrature (Q) signal components. In a traditional coherent receiver, the channel tap coefficients are estimated and then used to determine the symbol values from the received data. If a fractionally-spaced receiver is used, then the received data samples are viewed as multiple symbol-spaced data streams multiplexed together. Each symbol-spaced stream is modeled as shown above.
An example is given in the article authored by Y. Wan, Q. Liu and A. M. Sendyk, entitled “A fractionally-spaced maximum-likelihood sequence estimation receiver in a multipath fading environment” and published in ICASSP '92. Therein, differences between what was received and what was expected to be received are squared and summed to form a metric, which is minimized by the detected symbol values. For fractionally-spaced equalization, squared differences are still summed to determine the detected symbol values.
Such traditional approaches treat the transmit pulse shaping or filtering, the transmission medium, and the receive filtering together as one composite channel. If all the received samples have uncorrelated noise samples, then these traditional approaches are optimal. However, because the noise passes through the receive filter, it is bandlimited. Depending on the receive filter response and the sampling rate, the noise samples will be correlated. This occurs in symbol-spaced receivers when the receive filter is matched to the transmit filter and the composite response is not Nyquist. Moreover, partial response modulation schemes are intentionally designed with this property, so as to occupy a smaller bandwidth. This situation also arises in fractionally-spaced receivers when the receive filter is matched to the transmit filter and the filter bandwidth is such that the noise samples are correlated. Under these conditions, the traditional approaches are inaccurate.
One approach to solving this problem, which has been proposed for fractionally-spaced MLSE receivers, is to whiten the samples before traditional signal processing. See, for example, the articles authored by W. H. Sheen and G. Stüber, entitled “MLSE equalization and decoding for multipath-fading channels” published in IEEE Trans. Commun., Vol. 39, pp. 1455-1464, October 1991 and that authored by K. Hamied and G. L. Stüber, entitled “A fractionally spaced MLSE receiver” and published in ICC '95, Seattle, Wash., Jun. 18-22, 1995. However, if the receive filter is bandlimited or nearly so, then the whitening filter may be impossible or difficult to implement in practice. Also, whitening requires an additional filter in an operation, which adds complexity to the receiver.
Another solution is to use a wider receive filter, so that the noise samples are uncorrelated. A wider, “brick wall” filter is proposed in an article authored by G. M. Vachula and J. F. S. Hill, entitled “On optimal detection of band-limited PAM signals with excess bandwidth” which has been published in IEEE Trans. Commun., Vol. 29, pp. 886-890, June 1981. A practical, wider receive filter has been proposed in an article authored by K. Balachandran and J. B. Anderson, entitled “Receive filters for bandlimited intersymbol interference channels” which has been published in CISS '96, Princeton, N.J., March 1996 and in an article authored by H. Meyr, M. Oerder and A. Polydoros, entitled “On sampling rate, analog prefiltering, and sufficient statistics for digital receivers” and published in IEEE Trans. Commun., vol. 542, pp. 3208-3214, December 1994. However, this solution allows more noise to pass through the receive chain, which can cause saturation problems in, for example, the low noise amplifier when receiver dynamic range is limited. This is particularly troublesome when adjacent channel interference is present.
A similar problem occurs in spread spectrum systems. In a Rake receiver, the baseband samples r(k) correspond to correlations or despread values. These values are traditionally combined using estimates of the composite channel tap coefficients. As with nonspread systems, this is only optimal if the noise samples prior to despreading are uncorrelated. If the correlation spacing is chip-spaced and the chip pulse is not Nyquist or the correlation spacing is fractionally-spaced, then the noise on the despread values is correlated.
Thus, in both nonspread and spread communication systems, there is a need to improve receiver design to efficiently address the problem of noise correlation.
SUMMARY
The aforementioned problems, drawbacks and limitations of conventional signal processing techniques are overcome by the present invention by estimating the medium response separately from other effects, e.g., the effects associated with the receive and transmit filters, using knowledge of the pulse shaping. The medium response estimate is then used advantageously in the detection of information symbols.
For example, the medium response estimate can be used to provide the detection parameters used in a coherent detector to estimate the received information symbols. This application can be extended to array processors wherein a number of receive elements are provided to the receiver.
Moreover, the invention is applicable to both sprea
Bottomley Gregory E.
Molnar Karl
Ramesh Rajaram
Chin Stephen
Coats & Bennett PLLC
Ericsson Inc.
Ghayour Mohammad
LandOfFree
Method and apparatus for digital symbol detection using... 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 apparatus for digital symbol detection using..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for digital symbol detection using... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2473614