US 20020046382 A1 Abstract The present invention relates to a method and apparatus that uses cyclic redundancy check (CRC) bits to detect errors and correct particular four-bit error patterns in an information burst. The present invention is particularly suited for relatively simple systems that use differential encoding and either differential or coherent demodulation with no error correction capabilities, such as the Personal Handy Phone System (PHS), DECT, PACS and some digital cordless phones systems. The apparatus also adjusts its error-correction capability to accommodate different types of information bursts and channel qualities. The present invention may be used in a number of wireless, digital communication systems.
Claims(38) 1. An apparatus for detecting and correcting transmission errors using cyclic redundancy check bits in a communication system that has little or no error correction capability and uses differential encoding, said apparatus comprising:
a buffer unit for holding the bits of a received code word; a plurality of gates for deriving a syndrome from the received code word using a generator polynomial associated with said cyclic redundancy check; a syndrome register for holding the derived syndrome and cyclically shifting the derived syndrome one bit as each bit is read out of the buffer unit; an error detection module for detecting errors in the received code word; an error correction module that uses the derived syndrome to correct errors in the received code word by searching for a predetermined plurality of correctable error patterns in the derived syndrome and changing the corresponding bit of the received code word when a correctable error pattern is found after cyclic shifting, said error correction module activates when channel quality is above a user-programable threshold and deactivates when channel quality is below the user-programmable threshold; and a correction validation module for examining the syndrome register at the end of the decoding process and indicating whether the frame contains an uncorrectable error. 2. The apparatus of 3. The apparatus of 4. The apparatus of 5. The apparatus of 6. The apparatus of 7. The apparatus of 8. The apparatus of 9. The apparatus of 10. The apparatus of error metric from phase error;
difference in phase measured by an AFC loop;
a received signal strength indicator; and
a composite symbol detection error residuals measured for each frame.
11. The apparatus of 12. The apparatus of 13. The apparatus of 000X, where the 0's represent bits that are not in error and the Xs represent the bits that are in error.
00XX;
0X0X; and
X00X.
14. The apparatus of X000, where the 0's represent bits that are not in error and the Xs represent the bits that are in error.
XX00;
0X0X; and
X00X.
15. The apparatus of 16. The apparatus of 17. The apparatus of 18. The apparatus of 19. The apparatus of 20. The apparatus of 21. The apparatus of 22. The apparatus of 23. The apparatus of 24. The apparatus of 25. The apparatus of 26. The apparatus of 27. An apparatus for detecting and correcting errors using cyclic redundancy check bits in a communication system that uses differential coding, said apparatus comprising:
a memory for storing the bits of a received code word; a syndrome register for deriving a CRC syndrome from the received code word using a generator polynomial associated with the cyclic redundancy check and cyclically shifting the derived syndrome; an error detection and correction module to detect errors in the received code word and to correct errors in the received code word by searching for a predetermined plurality of correctable error patterns in the derived syndrome and changing the corresponding bit of the received code word when a correctable error pattern is found; and a correction validation module for examining the syndrome register at the end of the decoding process and indicating whether the received code word contains an uncorrectable frame error. 28. A method of detecting and correcting transmission error patterns using cyclic redundancy check bits in a communication system that uses differential encoding, said method comprising:
receiving a frame of encoded data from a transmitter across a channel; deriving a CRC syndrome for the received code word using a generator polynomial associated with said cyclic redundancy check; using the derived syndrome to determine if there are any errors in the information bits; determining whether the channel quality is above or below a certain user-programmable threshold; if channel quality is above the user-programmable threshold, perform error correction using the derived syndrome by cyclically shifting the derived syndrome and searching for a predetermined plurality of error patterns; examining the syndrome register at the end of the decoding process and indicating whether the received code word contains an uncorrectable frame error. 29. The method of 30. The method of 31. The method of 32. A communication system comprising:
a base station for transmitting information bursts, said base station uses differential encoding and adds cyclic redundancy check bits to the information bursts by using a generator polynomial for the purpose of error detection; a mobile unit which uses the same generator polynomial used by the base station to detect errors and correct errors without altering the encoding system used by the base station or adding more components to the base station, said errors are corrected by searching for a set of error patterns and changing the corresponding bit in the information burst when one of the error patterns is found. 33. A method of improving the performance of a communication system which uses differential encoding and generates cyclic redundancy check bits using a generator polynomial, said method comprising:
adding error correction capability to one or more mobile units which receive encoded information bursts from a base station, said error correction using the same CRC generator polynomial used by the base station. 34. A mobile receiver unit useful in a wireless communication system, that provides for no bit error correction and uses cycle redundancy checking (CRC) for error detection, the mobile receiver unit comprising:
an RF receiver to receive signals transmitted from a base station; a decoder in communication with said receiver to decode incoming signals and provide a series of frames containing data and CRC syndrome bits; an error detection and correction module to correct one of a plurality of predetermined error patterns in a single frame of information bits having a memory module to temporarily store each received frame, an error detection circuit to check for errors as the frame is shifted cyclically and an error correction circuit to correct errors wherein said mobile receiver unit uses only cyclic redundancy check syndromes for error correction. 35. The mobile receiver unit of 36. The mobile receiver unit of 37. The mobile receiver unit of 38. The mobile receiver unit of Description [0001] 1. Field of the Invention [0002] The present invention relates generally to wireless communication devices. Specifically, the present invention relates to the detection and correction of errors in received signals. [0003] 2. Brief Description of the Related Art [0004] In block coding for wireless digital communications, a transmitter transmits information bits with parity check bits added to the end of the information bits. Information bits may include control information (including initialization, synchronization, etc.), data, or voice message. The information bits and parity check bits together make up a code word or vector. At the receiving end, a receiver decodes the received code words and uses the parity check bits to detect errors in the received code word. Cyclic coding is one form of block coding. Shortened cyclic coding is an abbreviated form of cyclic coding. See [0005] Any decoder that tries to detect errors has a probability of misdetecting errors. As seen by the receiver, ‘misdetection’ includes at least two situations. If a frame contains an error and the decoder fails to detect this error (i.e. lets the frame pass for further processing as error-free), then there is a misdetection. If the decoder detects an error and miscorrects the error, then this is also a misdetection because the decoder ‘thinks’ it made a proper correction. Any decoder that attempts to correct errors increases the probability of misdetecting frame errors. This can degrade system performance. [0006] The present invention relates to a method and apparatus for detecting and correcting errors using CRC. The present invention is preferably used in communication systems that have error detection but no error correction capabilities. For example, the Personal Handy Phone System (PHS), a Japanese communication standard, has error detection but no error correction capability. See [0007] Currently, CRC is used to detect random or burst errors in DBPSK, DQPSK and π/4 DQPSK communication systems with either differential or coherent demodulation. CRC possesses a high degree of error detection capability compared to other forms of coding. In relatively simple systems like PHS, if the decoder detects an error, the decoder erases the information burst (e.g. for speech bits) or requests the transmitter to retransmit the same burst (e.g. for data bits). [0008] According to open literature, conventional CRC with a minimum distance of four can be used to correct a single-bit or a double-adjacent-bit error pattern in a single data frame. But if there are more errors than a single-bit or a double-adjacent-bit error pattern in a single frame, then CRC does not guarantee successful error correction. Once CRC is used to correct single-bit and double-adjacent bit errors, CRC quickly loses its high error detection capability. Current, practical communication systems do not use CRC to correct errors because correcting a single-bit or a double-adjacent-bit error pattern in a control or data frame does not justify the increased probability of misdetection associated with error correction. In other words, it is dangerous to sacrifice CRC's high error detection capability, particularly for control bursts, to correct limited error patterns. Control bursts require perfect or near-perfect transmission. As a result, where error correction is needed, a more sophisticated error correction system is often used, such as Reed-Soloman and convolutional coding. One problem with this type of error correction is that it is incompatible with the simpler system referenced above and it would add to the cost of the equipment. [0009] Another reason why practical communication systems do not use CRC to correct errors is that error correction becomes counterproductive when the channel is noisy, which causes multiple transmission errors in each frame or burst of information. Error correction using CRC in this case would increase the probability of creating more errors. Current industry standards and systems teach away from using CRC for error correction. [0010] The present invention recognizes that any modulation system that converts two bits to one symbol using differential encoding will cause double-adjacent symbol errors when each symbol is converted to two bits at the demodulation (decoding) end. The present invention further recognizes that these double-adjacent symbol errors correspond to a limited number of four-bit error patterns, and that these four-bit error patterns can be corrected. The present invention uses CRC to correct these four-bit error patterns without adding significant hardware and software. This is particularly advantageous in systems that have little or no error correction capabilities. [0011] By using CRC to correct errors, the present invention increases throughput (amount of information bits received and actually used) compared to current systems because the decoder in the present invention does not simply erase entire information bursts or requests the transmitter to retransmit the information bursts. The decoder in the present invention corrects errors in information bursts using CRC. For example, if 10 information bursts are sent in a system with no error correction and 3 out of 10 information bursts contain errors, then those 3 information bursts must be erased or retransmitted. Erasing information bits decreases throughput and degrades performance. Retransmission takes additional time, which also degrades performance. But 2 out of 3 of the information bursts with errors may contain error patterns which are correctable by the present invention. Thus, the present invention increases throughput without adding more redundant check bits to the information burst or adding significantly more circuitry to the decoder. [0012] In addition, the present invention recognizes that different types of information bursts (data, voice or control) in different channel conditions need different degrees of error detection capability. For example, high error detection is required for all control bursts. Because of the importance of a control burst, it has a very low tolerance for errors. Using CRC to correct errors may create additional errors if the channel quality is particularly poor. Thus, the use of CRC to correct errors should be done only when the signal is of a very high quality. In contrast, routine data bursts, representing voice data for example, can tolerate some additional errors which a decoder might make if it uses CRC to correct errors where the signal quality is somewhat lower. In one preferred embodiment of the present invention, error correction is turned on for voice bursts and is turned off for control bursts. [0013] Similarly, the present invention recognizes that when the channel quality is poor (noisy channels), error correction is likely to create more errors and misdetections, and thus deteriorate performance. But when channel quality is relatively good, error correction using CRC can significantly improve performance in systems that have little or no error correction capability. [0014] The present invention recognizes that a decoder may turn error correction using CRC on and off depending on the type of burst and channel quality. This makes the present invention more efficient when compared to current systems using a standard length CRC solely for error detection for both data and control bursts. For example, for a system that uses a standard 16-bit CRC, the present invention uses all 16 bits for both error detection and error correction. The amount of error correction varies, depending on the channel quality and the type of burst. In contrast, other systems use the same standard 16-bit CRC only for error detection, regardless of channel quality and type of burst. The present invention makes better use of the information provided by CRC and accomplishes much more with the same number of CRC bits, which in this example is 16. [0015] The present invention recognizes that the accuracy of channel quality estimation is important to apply CRC for error correction. The present invention uses a composite channel quality estimation which combines different channel quality estimates from various estimating methods. [0016] The present invention corrects random errors due to non-perfect timing effects in fading channels and resolves the error floor issue due to non-perfect radio frequency (RF) circuits. The present invention also minimizes the processing delay due to error corrections, minimizes the amount of added hardware needed to detect and correct errors without losing any bandwidth efficiency, and controls the misdetection probability of frame errors. [0017] Differential coding, cyclic codes and CRC are explained in further detail in [0018]FIG. 1 illustrates a transmitter and receiver in accordance with a preferred embodiment of the present invention. [0019]FIG. 2 illustrates the transmitter and receiver of FIG. 1 using an exemplary coding and modulation method. [0020]FIG. 3A illustrates an exemplary logic circuit for a decoder contained within the receiver of FIG. 2. [0021]FIG. 3B illustrated a preferred embodiment of the decoder of FIG. 3A. [0022]FIG. 4 is an exemplary logic circuit for the error pattern detection circuit of FIG. 3. [0023]FIG. 5 is a flow chart of the decoding method carried out by the circuit in FIG. 3. [0024]FIG. 6 is a flow chart of a frame error check carried out by the circuit in FIG. 3. [0025]FIG. 7 illustrates the relationship between probability of error misdetection with error correction and bit error rate. [0026]FIG. 8 illustrates the Bit Error Rate and Frame Error Rate versus signal to noise ratio between systems with and without error correction. [0027]FIG. 9A illustrates the differential coding regulations for a π/4 DQPSK system such as PHS. [0028]FIG. 9B illustrates a signal space diagram which shows the location of the four decoded absolute phases for a π/4 DQPSK system. [0029]FIG. 9C is a signal space diagram. [0030]FIG. 9D illustrates examples of three consecutive decoded absolute phases according to FIG. 9A in a noise-free channel, their possible variations in a noisy channel, and their corresponding correctable error patterns. [0031]FIG. 10A is another signal space diagram. [0032]FIG. 10B illustrate examples of three consecutive decoded absolute phases according to FIG. 10A in a noise-free channel, their possible variations in a noisy channel, and their corresponding correctable error patterns. [0033]FIG. 11A is another signal space diagram. [0034]FIG. 11B illustrate examples of three consecutive decoded absolute phases according to FIG. 11A in a noise-free channel, their possible variations in a noisy channel, and their corresponding correctable error patterns. [0035]FIG. 12A is another signal space diagram. [0036]FIG. 12B illustrate examples of three consecutive decoded absolute phases according to FIG. 12A in a noise-free channel, their possible variations in a noisy channel, and their corresponding correctable error patterns. [0037]FIG. 1 illustrates a method of packeting, coding, modulation, demodulation and decoding in accordance with the present invention. In general, a transmitter [0038] Specifically, the transmitter [0039]FIG. 2 illustrates the coding and decoding processes of FIG. 1 in more detail using an exemplary coding and modulation method. In general, a transmitter [0040] The transmitter [0041] When the channel [0042] In differential phase shift-keying (DPSK), demodulation of differentially encoded PSK signals does not require estimation of the carrier phase. The received signal is compared to the phase of the received signal from the preceding signalling interval. In other words, symbols are represented as the relative changes in phase rather than an absolute phase. [0043] Because of the nature of differential encoding and decoding, any single demodulation error may cause two adjacent symbol errors. These decision errors are carried over when the demodulator [0044] The amount and quality of error detection and correction depends on the type of CRC used. For example, a standard 16-bit CRC can detect all single and double bit errors and all errors as a result of odd parity in blocks of any practical length. As for error correction, a 16-bit CRC with a minimum distance of 4 can correct 1-bit errors and double-adjacent bit errors. [0045] In reviewing four-bit error patterns for a differential encoding system using π/4 QPSK and coherent demodulation, it was discovered that there are four four-bit error patterns covering most correctable error patterns which can be corrected using CRC. [0046] The four error patterns used in the preferred embodiment of the present invention will now be explained with reference to FIGS. [0047] In FIG. 9D, the first entry (0, 1, 2′) is an example of three consecutive decoded absolute phases in a noise-free channel. (0, 1, 2′) correspond to the absolute phases 0, π/4, and π/2 in FIG. 9C. The phase difference between each consecutive absolute phase is π/4, and the corresponding bits decoded are 0000 based on the table in FIG. 9A. If the same three demodulated absolute phases in a noisy channel are detected as (0, 1′, 2′), then the phase differences between the consecutive absolute phases are 3π/4 and −π/4 (the phase difference from 0 to 1′ is 3π/4, and the phase difference between 1′ and 2′ is −π/4). According to FIG. 9A, 3π/4 and −π/4 correspond to a bit pattern of 0110. Thus, the decoder [0048] In the present invention, only the last four (rightmost) bits of a 16-bit CRC shift register are used for error correction. The first twelve bits of the shift register are presumed to be zero during error correction. Thus, in the description below, an error pattern of 000X can be expressed as X, 0X, 000X or 000000000000000X for a 16-bit CRC. Similarly, an error pattern of 00XX can expressed as XX, 00XX or 00000000000000XX. [0049] Although there is a possibility that the second received absolute phase is 1′″ in a detected sequence (0, 1′″, 2′), the resulting bit error pattern would be 1111 or XXXX. This error pattern is an unlikely pattern, and wherever it appears, it indicates there are too many errors in one frame to correct. In the present invention, a channel quality indicator (not shown) would report that the frame containing (0, 1′″, 2′) is an uncorrectable frame and disable the error correction module [0050] In FIG. 9D, there are three more examples of three consecutive decoded absolute phases in a noise-free channel, represented as (0, 1, 2″), (0, 1, 2′″) and (0, 1, 2″″) and their possible variations in a noisy channel. The correctable bit error patterns corresponding to each example are listed in the righthand column. FIGS. 10A, 10B, [0051] Based on FIGS. [0052] Similarly, error patterns of 000X, 00XX, 0X0X, and X00X can be used for a differentially encoded QPSK system. Thus, for a differential encoding system using QPSK or π/4 QPSK and coherent demodulation, the most efficient, correctable four-bit error patterns are 000X, 00XX, 0X0X and X00X. For a differential encoding system using BPSK and coherent demodulation, the most efficient, correctable error patterns are found to be 0X and XX. [0053] In π/4 DQPSK systems with differential demodulation of signals, demodulation is different from coherent demodulation. Focusing on only the demodulated signal phases, the difference between two consecutive signal phases are decoded, rather than the differential decoding of two consecutive decoded phases in coherent demodulation. For differential demodulation, the present invention found the following correctable bit error patterns: X000, 0X00, 00X0, 000X, 0XX0, 0X0X, X0X0, and X00X. For error correction, these error patterns can be grouped as 000X, 00XX, 0X0X, X00X, similar to a coherent demodulation system described above. These four error patterns can be also used for a DQPSK case. For a BPSK system with differential encoded and differential demodulation, the correctable error patterns are found to be 0X and XX. [0054] In summary, the present invention recognized that for a BPSK system with differential encoding and either coherent or differential demodulation, the correctable bit error patterns are 0X and XX. For a QPSK or π/4 QPSK system with differential encoding and either coherent or differential demodulation, the error patterns for correction are 000X, 00XX, 0X0X and X00X. [0055] In a preferred embodiment, an error location module is added to reduce false correction probability and to decide if a particular error pattern needs correction. For example, for a π/4 QPSK system with differential encoding and either coherent or differential demodulation, if a bit error pattern of 00XX comes from one decoded symbol rather than two consecutive symbols, error correction should not be performed. For a π/4 QPSK system with differential encoding and coherent demodulation, if a bit error pattern of 000X does not come from the first or last decoded symbol, error correction should not be performed. [0056] The error detection and correction process will now be described with reference to FIG =1+ [0057] In accordance with cyclic encoding in systematic form (the rightmost k bits of each code word are unaltered information bits and the leftmost n−k bits are check bits), the encoder [0058] In shortened cyclic codes with 16 CRC bits, n is very large. In one embodiment, for example, n=2 [0059] Next, the encoder [0060] the transmitted code word or vector is 1( [0061] The transmitter [0062] At the receiver [0063] The method and apparatus described above may be used in a number of wireless, digital communication systems. The apparatus is particularly suited for a system that uses differential coding and either differential or coherent demodulation with error detection employing CRC. An exemplary use of the error correction system of the present invention is in a Personal Handy Phone System (PHS). The PHS system is a Time Division Duplexing (TDD) system, which uses one frequency for both transmission and reception on a time-sharing basis. See [0064] In the PHS system, there are 220 bits in a standard burst. A standard burst comprises a 2-bit start symbol (SS), a 6-bit or 62-bit preamble (PR), a 16-bit or 32-bit unique word (UW), 108 or 180 data bits (control signals or voice data), and a 16-bit CRC. For a control burst, k−L=108, n−L=108+16=124, and the total number of encoded burst (data plus CRC bits) is 124. For a traffic burst, k−L=180, n−L=180−16=196, and the total number of encoded burst (data plus CRC bits) is 196. Thus, n−k=16. The PHS system uses a series to parallel converter, differential coding with π/4 QPSK and a generator polynomial g(x)=1+x [0065] The error correction process of a preferred embodiment in an exemplary differential coding and coherent demodulation system (PHS) will now be described with reference to FIGS. 2 through 4. If the initial values of the syndrome shift register in the encoder [0066] According to the PHS specifications, the initial values of the syndrome shift register in the encoder [0067] For a shortened cyclic code, such as CRC, the encoder [0068] Here, b″(x) represents the contents of the syndrome register before any information bits enter the encoder [0069] The k−L information bits to be transmitted may be represented by [0070] The transmitted code word t(x) is then [0071] where t [0072] If no errors appear, then the received code word r(x) at the receiver [0073] where r [0074] The extra ‘L’ shifts can be eliminated, however, by modifying the connections of the syndrome register to divide x [0075] where ρ(x) is the remainder of x [0076] In a binary field with only 1's and 0's, [0077] and the above equation may be expressed as ρ( [0078] Thus, the decoder [0079] For a system such as PHS which sets all initial states of the syndrome register to 1, the left side of equation (3) can also be represented as ρ( =ρ( =ρ( =ρ( =ρ( [0080] because a binary field with only 1's and 0's has the following property: [0081] Combining equations (3) and (5) results in: [0082] Moving a3(x)g(x) to the right side yields: ρ( ρ( [0083] where a ρ( [0084] It is known by those of ordinary skill in the art that g(x) is always some factor of x [0085] where f(x) is some factor of (x ρ( [0086] Comparing equations (6) and (8) yields ρ( [0087] In an ideal channel with no transmission errors, [0088] Equation (9) suggests that in cases where all initial states are 1's, error detection is to check if all the syndrome registers are all 1's or not. If all the syndrome registers are 1's, then there are no errors. If the syndrome registers are not all 1's, then at least one error is detected. [0089] For an encoder with all initial states of the syndrome register set to 0's: [0090] and equation (9) becomes [0091] This means that if all the syndrome registers are all 0's, then there are no errors. If the syndrome registers are not all 0's, then at least one error is detected. Assuming no errors, [0092] where e(x) is the error vector. Thus, ρ( [0093] Substituting t(x) for r(x) in equation (3) yields, ρ( [0094] As shown in equation (11), g(x) is a factor of ρ(x)t(x). [0095] If there is some error e(x), then the quotient a1(x) in equation (1) is different: [0096] Because of the effect of the error e(x) on r(x), the new quotient a9(x) replaces a1( ρ( [0097] Multiplying every term in equation (10) by ρ(x) yields, ρ( [0098] Rearranging the terms of the above equation yields, ρ( [0099] Substituting equations (3) and (12) into equation (13) yields, ρ( [0100] Equation (14) is the standard form of a shortened cyclic code for an encoder with all initial states of the syndrome register set to 0's. The present invention recognizes and uses standard CRC decoding algorithm to apply equation (14) and the above equations for error correction. [0101] Current systems only use CRC for error detection because correcting a single-bit or a double-adjacent-bit error pattern in a control or data frame does not justify the increased probability of misdetection associated with error correction, as described above. In other words, it is dangerous to sacrifice CRC's high error detection capability, particularly for control bursts, to correct limited error patterns. Control bursts require perfect or near-perfect transmission. Thus, the present invention increases throughput (amount of information bits used) by using CRC for error correction without adding significantly more circuitry. The present invention is particularly suited for relatively simple systems with no other error correction capability, such as PHS (a Japanese standard), DECT (a European standard), PACS (a United States standard), and some digital cordless phones. Currently, these systems only use CRC for error detection. More complex systems use additional hardware and software for error correction methods, such as Reed Soloman. [0102] For a system such as the PHS system with all initial states of the syndrome register set to 1's, the received code word with some error e(x) can be expressed as, [0103] From equation (4), ρ( =ρ( [0104] Using equation (5), this can be expressed as, ρ( [0105] Substituting equations (8) into the above equation yields, ρ( [0106] Rewriting equation (16) and substituting equation (12) yields, ρ( =[a =[ = [0107] Let s [0108] Then equation (17) becomes ρ( [0109] Equation (19) is in the standard form of a shortened cyclic code as shown by equation (14). Because equation (19) is in the standard form of a shortened cyclic code, the decoder [0110]FIG. 3A illustrates a preferred embodiment for the decoder [0111] The purpose of gates g [0112] The initial frame error indicator [0113] Although the initial frame error indicator circuit [0114]FIG. 3B illustrates another preferred embodiment of the decoder [0115] The use and operation of the decoder [0116] In block [0117] To detect errors, the initial frame error indicator module [0118] In block [0119] In a preferred embodiment, channel quality is determined immediately after demodulation and simultaneously with error detection. If the channel quality is too poor, i.e. there is a significant amount of noise or interference in the signal which causes multiple transmission errors in a single burst, then error correction is likely to create more errors and misdetections, and thus deteriorate performance. As shown in block [0120] But when channel quality is relatively good, error correction using CRC can significantly improve performance in relatively systems that have little or no error correction capability. The present invention recognizes that a decoder [0121] Methods to measure channel quality (or SNR) include (1) received signal strength indicator (RSSI), (2) error metric from phase error, (3) difference in phase measured by an AFC loop or (4) composite symbol detection error residuals measured for the frame or from previous bursts. [0122] In a preferred embodiment, the present invention measures channel quality by using more than one of the channel quality indicators listed above and averaging the estimated SNR generated by each channel quality indicator. This is called a composite channel quality or signal quality. For example, the device may derive the SNR based on the difference in phase of an AFC loop and the SNR based on error metric from phase error. The device then averages the two SNRs to provide a more accurate SNR. If the channel quality indicators are uncorrelated (independent, i.e. measuring and calculating one channel quality indicator does not depend on measurements for another channel quality indicator), then taking the average SNR reduces variance and provides a much more accurate channel quality indication. This makes error correction more efficient because the device will activate error correction at a SNR level where the information bursts do not have multiple errors, and error correction using CRC does not create more errors and misdetections. The present device deactivates error correction using CRC as soon as the SNR drops below this SNR level. [0123] In a preferred embodiment, the channel quality threshold is programmed to be higher for bursts containing control information (e.g. synchronization) than for bursts containing traffic information (voice or data). In another preferred embodiment, instead of a higher threshold, the decoder [0124] If the decoder [0125] The decoder [0126] When the syndrome register [0127] Those of ordinary skill in the art will appreciate that the error detection module of FIGS. 3A, 3B and [0128] As shown in block [0129] In FIG. 3B, the syndrome register [0130]FIG. 6 is a flow chart of a frame correction validation check carried out by the correction validation module [0131] If the frame contains one or more uncorrectable errors, then the contents of the syndrome register [0132] In block [0133] In blocks [0134]FIG. 8 illustrates the signal to noise ratio between systems with and without error correction in AWGN channels. As depicted in FIG. 8, the BER and FER improve significantly if the decoder [0135] Thus, the present invention uses CRC to correct random errors due to non-perfect timing effects in fading channels, among other types of errors. The present invention also minimizes the processing delay due to error corrections, minimizes the amount of added hardware needed to detect and correct errors without losing any bandwidth efficiency, and controls the misdetection probability of frame errors. [0136] While embodiments and applications of this invention have been shown and described, it will be apparent to those skilled in the art that various modifications are possible without departing from the scope of the invention. It is, therefore, to be understood that within the scope of the appended claims, this invention may be practiced otherwise than as specifically described. Referenced by
Classifications
Legal Events
Rotate |