Decision directed DC offset removal
A method and apparatus for decoding a baseband signal of a radio signal removes, from the baseband signal, low-frequency and long-term noise that increases the possibility of decoding errors. The removal of low-frequency and long-term noise is performed by accumulating differences between the actual signal levels of the baseband signal and the expected signal levels for the baseband signal and subtracting the accumulated difference from the baseband signal before decoding. In one scheme, the baseband signal contains a predetermined training sequence of signal levels, where the differences between the actual signal levels of the baseband signal and the expected signal levels for the predetermined training sequence are accumulated. At the end of the training sequence, the accumulated training sequence difference is used as the accumulated difference and subtracted from the baseband signal, thereby providing stable operation for decoding signal levels that follow the training sequence.
1. A method comprising:
associating each of a plurality of expected values with a value range;
determining that a corrected signal falls within the value range of a particular expected value of the plurality of expected values;
deriving an error value by subtracting the particular expected value from the corrected signal; and
adding the error value to an accumulated error value;
limiting the accumulated error value to a particular range that is based on a minimum distance between the plurality of expected values;
wherein the corrected signal is derived by subtracting the accumulated error value from a received signal; and
wherein the received signal is derived from a radio signal.
2. The method of claim 1 , wherein the received signal is a baseband signal of the radio signal.
3. The method of claim 1 , wherein:
the received signal is an output of a high-pass filter; and
an input of the high-pass filter is a baseband signal of the radio signal.
4. The method of claim 1 , further comprising:
determining whether a signal level of the received signal corresponds to a signal level in a training sequence;
wherein the training sequence is a predetermined sequence of signal levels.
5. The method of claim 4 , further comprising:
in response to determining that a signal level of the received signal corresponds to a signal level in the training sequence, performing the steps of:
deriving a training sequence error value by subtracting the signal level of the received signal from a corresponding signal level in the training sequence; and
adding the training sequence error value to an accumulated training sequence error value;
in response to determining that a signal level of the received signal does not correspond to a signal level in the training sequence, performing the step of:
using the accumulated training sequence error value as the accumulated error value.
6. The method of claim 4 , wherein an average of the signal levels in the training sequence is zero.
7. The method of claim 4 , wherein the step of determining whether a signal level of the received signal corresponds to a signal level in a training sequence comprises determining whether a predetermined amount of time has elapsed since a previous determination that a signal level of the received signal corresponds to a signal level in the training sequence was made.
8. An apparatus comprising:
a comparator that:
associates each of a plurality of expected values with a value range; and
determines that a corrected signal falls within the value range of a particular expected value of the plurality of expected values;
a first subtractor that subtracts the particular expected value from the corrected signal to derive an error value;
an accumulator that adds the error value to an accumulated error value;
a limiter which limits the accumulated error value to a particular range that is based on a minimum distance between the plurality of expected values; and
a second subtractor that subtracts the accumulated error value from a received signal to derive the corrected signal;
wherein the received signal is derived from a radio signal.
9. The apparatus of claim 8 , wherein the received signal is a baseband signal of the radio signal.
10. The apparatus of claim 8 , wherein:
the received signal is an output of a high-pass filter; and
an input of the high-pass filter is a baseband signal of the radio signal.
11. The apparatus of claim 8 , further comprising:
a training sequence detector that determines whether a signal level of the received signal corresponds to a signal level in a training sequence;
wherein the training sequence is a predetermined sequence of signal levels.
12. The apparatus of claim 11 , further comprising:
a training sequence comparator that subtracts the signal level of the received signal from a corresponding signal level in the training sequence to derive a training sequence error value;
a training sequence error accumulator that adds the training sequence error value to an accumulated training sequence error value;
wherein the second subtractor, in response to the training sequence detector determining that a signal level of the received signal does not correspond to a signal level in a training sequence, subtracts the accumulated training sequence error value to derive the corrected signal.
13. The apparatus of claim 11 , wherein an average of the signal levels in the training sequence is zero.
14. The apparatus of claim 11 , wherein the training sequence detector determines whether a signal level of the received signal corresponds to a signal level in a training sequence by:
determining whether a predetermined amount of time has elapsed since the training sequence detector previously determined that a signal level of the received signal corresponds to a signal level in the training sequence.
15. The method of claim 1 , wherein each value range is unique.
16. The method of claim 15 , wherein combining the value ranges associated with the expected values forms a continuous value range.
17. The apparatus of claim 8 , wherein each value range is unique.
18. The apparatus of claim 17 , wherein combining the value ranges associated with the expected values forms a continuous value range.