Patent US5428644A successfully refreshed from Google Patents.
US5428644A
Expired / Lapsed
US
⏱️ 0.0 years remaining
Adaptive detection method and detector for quantized signals
Assignee: Nokia Oyj
•
Filing: 1993-06-10
•
Granted: 1995-06-27
•
Expires: 2013-06-10
Abstract
Abstract The invention relates to an adaptive detection method and detector for detecting signal states of a received quantized signal using a set of reference vectors stored in a memory (11). According to the invention, the detector comprises sampling means (4) for sampling the received signal in a sliding time window; means (2) using a self-organizing map for calculating a first corrected set of reference vectors on the basis of the stored set of reference vectors used in the detection and samples of actual signal states received in the time window; means (3) employing the K-means method or the like for calculating a second corrected set of reference vectors on the basis of the first corrected set of reference vectors and the samples of the actual signal states; and means (9) for updating the set of reference vectors stored in the memory means (11) by the second corrected set of reference vectors.
Inventors
- 👤 Teuvo Kohonen
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-10-28 23:37
CPC Classifications
H; H04; H04L; H04L27/00; H04L27/32; H04L27/34; H04L27/38; H04L25/00; H04L25/02; H04L25/06; H04L25/061; H04L25/062
Claims Analysis
Total Claims: 10
Claim 1
Independent Claim
339 words
An adaptive detection method for quantized signals with a finite number of signal states, said method comprising the steps of receiving a quantized signal, detecting the signal states of the received signal using a set of reference vectors stored in a memory, updating the stored set of reference vectors on the basis of samples of the actual signal states of the received signals, the updating of the reference vectors comprising a) calculating a first corrected set of reference vectors by a method employing a self-organizing map, on the basis of a stored set of reference vectors used in the detection and samples taken from actual received signal states in a predetermined time window, the calculation step comprising the sub-steps of a1) using as the starting point the present stored set of reference vectors used in the detection, a2) collecting a sample set for each reference vector, said sample set containing all the signal samples whose closest reference vector belongs to the set of neighbors of the first corrected set of reference vectors, a3) calculating for each reference vector the average of the collected sample set, a4) forming a first corrected set of reference vectors from said calculating average, of the collected sample sets, a5) repeating sub-steps a2), a3) and a4) for the same sample set with a desired number of iteration cycles using as the starting point in sub-step a2) the first corrected set of reference vectors formed during the previous iteration cycle, b) calculating a second corrected set of reference vectors by utilizing a K-means type method on the basis of the first corrected set of reference vectors and said samples of the actual signal states, c) updating the stored set of reference vectors used in the detection by said second corrected set of reference vectors, d) moving the time window in the time domain for sampling a new sample set from actual signal states of the received signal, and e) repeating process steps a), b), c) and d) for each new time window defined at step d).
Claim 2
Dependent Claim
339 words
An adaptive detection method for quantized signals with a finite number of signal states, said method comprising the steps of receiving a quantized signal, detecting the signal states of the received signal using a set of reference vectors stored in a memory, updating the stored set of reference vectors on the basis of samples of the actual signal states of the received signals, the updating of the reference vectors comprising a) calculating a first corrected set of reference vectors by a method employing a self-organizing map, on the basis of a stored set of reference vectors used in the detection and samples taken from actual received signal states in a predetermined time window, the calculation step comprising the sub-steps of a1) using as the starting point the present stored set of reference vectors used in the detection, a2) collecting a sample set for each reference vector, said sample set containing all the signal samples whose closest reference vector belongs to the set of neighbors of the first corrected set of reference vectors, a3) calculating for each reference vector the average of the collected sample set, a4) forming a first corrected set of reference vectors from said calculating average, of the collected sample sets, a5) repeating sub-steps a2), a3) and a4) for the same sample set with a desired number of iteration cycles using as the starting point in sub-step a2) the first corrected set of reference vectors formed during the previous iteration cycle, b) calculating a second corrected set of reference vectors by utilizing a K-means type method on the basis of the first corrected set of reference vectors and said samples of the actual signal states, c) updating the stored set of reference vectors used in the detection by said second corrected set of reference vectors, d) moving the time window in the time domain for sampling a new sample set from actual signal states of the received signal, and e) repeating process steps a), b), c) and d) for each new time window defined at step d).
Claim 3
Dependent Claim
60 words
A method according to claim 1 wherein process step a) the calculation of the first corrected set of reference vectors comprises at least two iteration cycles with a method employing a self-organizing map, and that in process step b) the calculation of the second corrected set of reference vectors comprises at least two iteration cycles with the K-means type method.
Claim 4
Dependent Claim
77 words
A method according to claim 1 or 2, wherein process step a2) comprises weighting of each sample by a conditional weight coefficient in such a manner that the weight coefficient is greater than one for the samples that are closest to any of the peripheral reference vectors in the set of reference vectors and fall outside the set of reference vectors, and that in process a3) said averages of sample sets are calculated using weighted signal samples.
Claim 5
Dependent Claim
31 words
A method according to claim 1, wherein in step b) the calculation of the second corrected set of reference vectors comprises at least two iteration cycles with the K-means type method.
Claim 6
Dependent Claim
74 words
A method according to claim 1, wherein sub-step a2) comprises weighting of each sample by a conditional weight coefficient in such a manner that the weight coefficient is greater than one for the samples that are closest to any of the peripheral reference vectors in the set of reference vectors and fall outside the set of reference vectors, and that in step a3) said averages of sample sets are calculated using weighted signal samples.
Claim 7
Dependent Claim
320 words
An adaptive detection method for quantized signals with a finite number of signal states, said method comprising the steps of receiving a quantized signal, detecting the signal states of the received signal using a set of reference vectors stored in a memory, updating the stored set of reference vectors on the basis of sample of the actual signal states of the received signals, the updating of the reference vectors comprising a) calculating a first corrected set of reference vectors by a method employing a self-organizing map, on the basis of a stored set of reference vectors used in the detection and samples taken from actual received signal states in a predetermined time window, b) calculating a second corrected set of reference vectors by utilizing a K-means type method on the basis of the first corrected set of reference vectors and said samples of the actual signal states, the calculation step comprising the sub-steps of b1) using as the starting point the first corrected set of reference vectors formed in step a), b2) providing each signal sample with an index of the closest reference vector in the set of reference vectors, b3) calculating for each index the average of the signal samples provided with the same index, b4) forming the second corrected set of reference vectors from said calculated average, b5) repeating sub-steps b2), b3) and b4) for the same sample set with a desired number of iteration cycles using as the starting point in sub-step b2) the second corrected set of reference vectors formed during the previous iteration cycle, c) updating the stored set of reference vectors used in the detection by said second corrected set of reference vectors, d) moving the time window in the time domain for sampling a new sample set from actual signal states of the received signal, and e) repeating steps a), b), c) and d) for each new time window defined at step d).
Claim 8
Dependent Claim
320 words
An adaptive detection method for quantized signals with a finite number of signal states, said method comprising the steps of receiving a quantized signal, detecting the signal states of the received signal using a set of reference vectors stored in a memory, updating the stored set of reference vectors on the basis of sample of the actual signal states of the received signals, the updating of the reference vectors comprising a) calculating a first corrected set of reference vectors by a method employing a self-organizing map, on the basis of a stored set of reference vectors used in the detection and samples taken from actual received signal states in a predetermined time window, b) calculating a second corrected set of reference vectors by utilizing a K-means type method on the basis of the first corrected set of reference vectors and said samples of the actual signal states, the calculation step comprising the sub-steps of b1) using as the starting point the first corrected set of reference vectors formed in step a), b2) providing each signal sample with an index of the closest reference vector in the set of reference vectors, b3) calculating for each index the average of the signal samples provided with the same index, b4) forming the second corrected set of reference vectors from said calculated average, b5) repeating sub-steps b2), b3) and b4) for the same sample set with a desired number of iteration cycles using as the starting point in sub-step b2) the second corrected set of reference vectors formed during the previous iteration cycle, c) updating the stored set of reference vectors used in the detection by said second corrected set of reference vectors, d) moving the time window in the time domain for sampling a new sample set from actual signal states of the received signal, and e) repeating steps a), b), c) and d) for each new time window defined at step d).
Claim 9
Dependent Claim
139 words
A method according to claim 6, wherein process step a) comprises the steps of a1) using as the starting point the present stored set of reference vectors used in the detection, a2) collecting a sample set for each reference vector, said sample set containing all the signal samples whose closest reference vector belongs to the set of neighbors of the first corrected set of reference vectors, a3) calculating for each reference vector the average of the collected sample set, a4) forming a first corrected set of reference vectors form said calculated average, of the collected sample sets, a5) repeating process steps a2), a3) and a4) for the same sample set with a desired number of iteration cycles using as the starting point in process step a2) the first corrected set of reference vectors formed during the previous iteration cycle.
Claim 10
Dependent Claim
34 words
A method according to claim 6, wherein in process step a) the calculation of the first corrected set of reference vectors comprises at least two iteration cycles with a method employing a self-organizing map.
References Cited (Backward Citations)
USPTO Annuity & Maintenance Windows
| Annuity Stage | Window Opens | Payment Deadline | Surcharge Period Until | Estimated Fee | Status |
|---|---|---|---|---|---|
| 3.5 Years (Year 4) | 1998-06-26 | 1998-12-26 | 1999-06-27 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 2002-06-26 | 2002-12-26 | 2003-06-27 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 2006-06-26 | 2006-12-26 | 2007-06-27 | $7700.00 | Window Closed / Lapsed |