Error detection/correction and fault detection/recovery – Pulse or data error handling – Digital data error correction
Reexamination Certificate
1998-08-31
2001-08-07
Tu, Christine T. (Department: 2784)
Error detection/correction and fault detection/recovery
Pulse or data error handling
Digital data error correction
C714S795000
Reexamination Certificate
active
06272660
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to data error screening and, in particular embodiments, to screening for errors in systems in which an error mitigation signal may be produced.
In a typical data transmission system, a stream of data bits is encoded in a transmitter and transmitted over a communications channel that exhibits noise, distortion or other impairments. In the receiver, a corrupted version of what was transmitted is operated on by a decoder to generate an output sequence of bits that is determined to be that which most likely corresponds to the stream of bits that was encoded in the transmitter.
In some such systems, the decoded output sequence undergoes some type of “analysis” resulting either in an acceptance of the decoded output sequence as being correct or in an indication of a potential decoding error and hence a rejection of the decoded output sequence. A typical such analysis involves error detection and/or correction using codes well known in the art, such as parity check information that was included in the encoded stream of data bits. If the decoded output sequence corresponding to the transmitted encoded data proves to have the correct parity, it is released for further processing. Otherwise an error flag is released, responsive to which the system may either (a) request a re-transmission of the encoded data or (b) initiate error mitigation routines. In audio data transmission systems, for example, the error mitigation routines may substitute a so-called error mitigation signal for the audio signal corresponding to the decoded sequence. This approach is analogous to a long-standing practice in video transmission systems. See, for example,
The Digital Signal Processing Handbook,
CRC Press, 1998, ch. 56.
The decoder may be capable of identifying more than one output sequence corresponding to the encoded stream of data bits. An example of such a decoder is a so-called List Viterbi decoder which utilizes a List Viterbi algorithm to identify the L best candidate decodings corresponding to the encoded stream of bits. The aforementioned analysis, then, may involve the determination of which, if any, decoded output sequence should be selected for further processing.
For certain transmission systems, such as high quality audio systems, the analysis step is particularly important. Errors in a decoded sequence that would be noticeable to a listener may under certain circumstances- for example, if the errors are few in number and isolated-be mitigated upon being detected. On the other hand, errors in a decoded sequence that are undetected during the analysis step and thus are present in the decoded sequence as it is being further processed by an audio decoder-may be very significant, resulting, for example, in a listener hearing clicks during a pause in a musical selection. In such systems, then, it is desirable to be able to further screen decoded sequences for errors which were undetected in the analysis step.
SUMMARY OF THE INVENTION
In prior art arrangements as just described, the decision as to whether a decoded sequence is acceptable is made by determining the likelihood of significant errors (hereinafter referred to as “error screening” in the decoded sequence based solely on criteria intrinsic to the decoded sequence itself, such as its satisfaction of a parity check. That is, for example, the prior art analysis step is solely a function of the decoded sequence (including information derivable from that sequence). In accordance with the present invention, by contrast, error screening of a decoded sequence is based on at least one criterion that is both extrinsic and intrinsic to the decoded sequence currently being screened. That is, for example, error screening is a function of intrinsic information, i.e., the decoded sequence and/or a function of the decoded sequence, and extrinsic information, i.e., information exclusive of that sequence. This approach, we have found, is likely to lead to an increased number of errors being detected than that detected in prior art arrangements, thereby improving, for example in audio data transmission systems, the overall quality of the audio signal.
More particularly, the error screening criterion may be a comparison between a signal corresponding to the decoded sequence and a so-called error mitigation signal. As is well known, an error mitigation signal is a signal generated in the receiver and used as an estimate of the signal corresponding to the decoded sequence that had been determined to be unacceptable. The error mitigation signal is typically generated from decoded sequences (or other information) other than the decoded sequence to which it is being compared. Thus the comparison between the signal corresponding to the decoded sequence with the error mitigation signal meets the criterion above as being both extrinsic and intrinsic to that decoded sequence.
As a consequence of the comparison, a decision is made as to whether the decoded sequence is applied to an output, illustratively released or used for further processing in the receiver. That is, for example, depending on the results of the comparison, either the decoded sequence is accepted and the signal corresponding to the decoded sequence is applied to an output, or the error mitigation signal is applied. In preferred embodiments, error screening is performed only when most advantageous to do so, for example, if a particular condition indicative of an increased likelihood of undetected errors exists.
There are prior art systems which determine whether a decoded sequence is acceptable. However, that determination is made in these systems by using criteria solely extrinsic to the sequence. That is, the determination is based on information exclusive of the decoded sequence. For example, the receiver may check channel conditions at the time of transmission and may determine that conditions were such that the received symbols are most likely too error-flawed to warrant being decoded. However, since, in fact, no decoding or screening occurs, these systems do not screen for errors contained in a decoded sequence and certainly do not constitute systems in which, per the present invention, such screening is based on at least one criterion that is intrinsic, as well as extrinsic, to that sequence.
The present invention is particularly advantageous in systems, such as audio data transmission systems, in which a List Viterbi decoder is utilized. More specifically, we have realized that while use of a List Viterbi algorithm significantly reduces the so-called error flag rate for digital audio applications, the lower error flag rate motivates the use of such applications in noisier channels where the probability of errors that are undetected by prior art analysis techniques increases. The problem of errors undetected by the Viterbi decoding may be dealt with by including a parity check code, for example, a Cyclic Redundancy Check (CRC) code. However, there still remains errors undetected by the parity check code, the occurrence of which can be mitigated against by increasing the number of bits in the existing parity check code. Assuming that the channel bandwidth is fixed, this technique leads to a decreased data rate. In contrast, the present invention advantageously lowers the rate of undetected errors that may have a significant impact on decoded signal quality without decreasing the data rate, while engendering only a relatively small amount of increased processing complexity and delay.
In a number of the embodiments described hereinbelow, error screening in a decoded sequence occurs after a preliminary analysis as a result of which that decoded sequence was selected for further processing. However, error screening may in fact also be used in the selection of the decoded sequence in the first instance. For example, in systems utilizing List Viterbi decoding, techniques using principles of the invention may be used to select which of the candidate decodings identified by List Viterbi decoding should be released for further processing.
RE
Chen Brian
Sundberg Carl-Erik Wilhelm
Agere Systems Guardian Corp.
Monteleone Geraldine
Tu Christine T.
LandOfFree
Screening for errors in data transmission systems does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Screening for errors in data transmission systems, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Screening for errors in data transmission systems will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2486352