Spur mitigation techniques
View Patent ↗Spurs cause significant problems with signal detecting, amplifier gain adjustment, and signal decoding. Various techniques can be used to mitigate the effects of spurs on a received signal. Generally, these techniques work by either canceling or ignoring the spurs. For example, a pilot mask can be used to ignore pilot information in one or more sub-channels. A Viterbi mask can determine the weighting given to bits in a sub-channel based on spur and data rate information. Channel interpolation can compute a pseudo channel estimate for a sub-channel known to have a spur location can be computed by interpolating the channel estimates of adjacent good sub-channels. Filtering of the received signal using a low-pass filter, a growing box filter, or a low-pass filter with self-correlation can be used to cancel a spur.
1. A method for canceling a spur from a signal, the method comprising:
rotating the signal to generate a first rotated signal;
performing a filtering computation based on the first rotated signal to generate a filtered signal;
rotating the filtered signal to generate a second rotated signal; and
subtracting the second rotated signal from the signal.
2. The method of claim 1 , wherein rotating the signal is performed such that a frequency of the spur is at 0 frequency.
3. The method of claim 1 , wherein rotating the signal is performed such that a frequency of the spur is at DC.
4. The method of claim 1 , wherein rotating the signal is expressed by:
z[n]=y[n]e −j2πf s nT
wherein y[n] is a model without noise, f s is a frequency of the spur, n is an index of a particular sample of the signal, and T is a sampling period.
5. The method of claim 4 , wherein performing the filtering computation is expressed by:
A
^
=
∑
k
=
0
l
-
1
h
[
k
]
z
[
n
-
k
]
wherein h[k] is a filter coefficient, and l is a filter length.
6. The method of claim 1 , wherein performing the filtering computation is performed by a low-pass filter.
7. The method of claim 1 , wherein rotating the filtered signal to generate the second rotated signal includes rotating the filtered signal in a direction opposite of that used when rotating the signal.
8. The method of claim 1 , wherein rotating the filtered signal to generate the second rotated signal includes returning the spur to its original frequency.
9. The method of claim 1 , wherein performing the filtering computation includes:
setting a sample set size;
computing a cumulative sum for the sample set over time;
when the sample set size is reached, then adding the cumulative sum to a total sum and resetting the cumulative sum to zero;
computing an estimated spur value by dividing the total sum by a total number of samples, wherein the estimated spur value is provided as the filtered signal; and
periodically increasing the sample set size over time.
10. The method of claim 9 , wherein the cumulative sum is performed by a growing box filter.
11. The method of claim 9 , wherein the cumulative sum is performed by an expandable low-pass filter.
12. The method of claim 11 , wherein the expandable low-pass filter uses sampled data up to a current symbol.
13. The method of claim 12 , wherein a sample size during each sample period is increased to provide doubling of total sample size.
14. The method of claim 12 , wherein a filter length is set to a maximum length to limit latency of computing the cumulative sum.
15. The method of claim 12 , wherein a filter length is increased until a substantially similar value is computed during two consecutive sampling periods.
16. The method of claim 9 , wherein if the cumulative sum is denoted by cs[n] and the total sum is denoted by ts[n], then adding the cumulative sum to the total sum and resetting the cumulative sum to zero occurs when n is a power of 2.
17. The method of claim 16 , wherein a first accumulator generates the cumulative sum and a second accumulator generates the total sum.
18. A method for canceling a spur from a signal, the method comprising:
rotating the signal to generate a first rotated signal;
performing a filtering computation based on the first rotated signal to generate a filtered signal;
rotating the filtered signal to generate a second rotated signal;
subtracting the second rotated signal from the signal to generate a modified signal; and
after subtracting, computing self-correlation of the modified signal.
19. The method of claim 18 , wherein rotating the signal is performed such that a frequency of the spur is at 0 frequency.
20. The method of claim 18 , wherein rotating the signal is performed such that a frequency of the spur is at DC.
21. The method of claim 18 , wherein rotating the signal is expressed by:
z[n]=y[n]e −j2πf s nT
wherein y[n] is a model without noise, f s is a frequency of the spur, n is an index of a particular sample of the signal, and T is a sampling period.
22. The method of claim 21 , wherein performing the filtering computation is expressed by:
A
^
=
∑
k
=
0
l
-
1
h
[
k
]
z
[
n
-
k
]
wherein h[k] is a filter coefficient, and l is a filter length.
23. The method of claim 18 , wherein performing the filtering computation is performed by a low-pass filter.
24. The method of claim 18 , wherein rotating the filtered signal to generate the second rotated signal includes rotating the filtered signal in a direction opposite of that used when rotating the signal.
25. The method of claim 18 , wherein rotating the filtered signal to generate the second rotated signal includes returning the spur to its original frequency.
26. The method of claim 18 , wherein performing the filtering computation includes:
setting a sample set size;
computing a cumulative sum for the sample set over time;
when the sample set size is reached, then adding the cumulative sum to a total sum and resetting the cumulative sum to zero;
computing an estimated spur value by dividing the total sum by a total number of samples, wherein the estimated spur value is provided as the filtered signal; and
periodically increasing the sample set size over time.
27. The method of claim 26 , wherein the cumulative sum is performed by a growing box filter.
28. The method of claim 26 , wherein the cumulative sum is performed by an expandable low-pass filter.
29. The method of claim 28 , wherein the expandable low-pass filter uses sampled data up to a current symbol.
30. The method of claim 29 , wherein a sample size during each sample period is increased to provide doubling of total sample size.
31. The method of claim 30 , wherein a filter length is set to a maximum length to limit latency of computing the cumulative sum.
32. The method of claim 30 , wherein a filter length is increased until a substantially similar value is computed during two consecutive sampling periods.
33. The method of claim 26 , wherein if the cumulative sum is denoted by cs[n] and the total sum is denoted by ts[n], then adding the cumulative sum to the total sum and resetting the cumulative sum to zero occurs when n is a power of 2.
34. The method of claim 33 , wherein a first accumulator generates the cumulative sum and a second accumulator generates the total sum.
35. A method for canceling a spur from self-correlation of a signal, the method comprising:
rotating the signal to generate a first rotated signal;
performing a filtering computation based on the first rotated signal to generate a filtered signal;
rotating the filtered signal to generate a second rotated signal, which represents a spur effect;
computing self-correlation of the signal to generate a modified signal; and
subtracting the spur effect from the modified signal.
36. The method of claim 35 , wherein computing self-correlation includes storing 32 samples.
37. The method of claim 35 , wherein rotating the signal is performed such that a frequency of the spur is at 0 frequency.
38. The method of claim 35 , wherein rotating the signal is performed such that a frequency of the spur is at DC.
39. The method of claim 35 , wherein rotating the signal is expressed by:
z[n]=y[n]e −j2πf s nT
wherein y[n] is a model without noise, f s is a frequency of the spur, n is an index of a particular sample of the signal, and T is a sampling period.
40. The method of claim 39 , wherein performing the filtering computation is expressed by:
A
^
=
∑
k
=
0
l
-
1
h
[
k
]
z
[
n
-
k
]
wherein h[k] is a filter coefficient, and l is a filter length.
41. The method of claim 35 , wherein performing the filtering computation is performed by a low-pass filter.
42. The method of claim 35 , wherein rotating the filtered signal to generate the second rotated signal includes rotating the filtered signal in a direction opposite of that used when rotating the signal.
43. The method of claim 35 , wherein rotating the filtered signal to generate the second rotated signal includes returning the spur to its original frequency.
44. The method of claim 35 , wherein performing the filtering computation includes:
setting a sample set size;
computing a cumulative sum for the sample set over time;
when the sample set size is reached, then adding the cumulative sum to a total sum and resetting the cumulative sum to zero;
computing an estimated spur value by dividing the total sum by a total number of samples, wherein the estimated spur value is provided as the filtered signal; and
periodically increasing the sample set size over time.
45. The method of claim 44 , wherein the cumulative sum is performed by a growing box filter.
46. The method of claim 44 , wherein the cumulative sum is performed by an expandable low-pass filter.
47. The method of claim 46 , wherein the expandable low-pass filter uses sampled data up to a current symbol.
48. The method of claim 47 , wherein a sample size during each sample period is increased to provide doubling of total sample size.
49. The method of claim 47 , wherein a filter length is set to a maximum length to limit latency of computing the cumulative sum.
50. The method of claim 47 , wherein a filter length is increased until a substantially similar value is computed during two consecutive sampling periods.
51. The method of claim 44 , wherein if the cumulative sum is denoted by cs[n] and the total sum is denoted by ts[n], then adding the cumulative sum to the total sum and resetting the cumulative sum to zero occurs when n is a power of 2.
52. The method of claim 51 , wherein a first accumulator generates the cumulative sum and a second accumulator generates the total sum.