The present disclosure relates generally to wireless communications systems receivers and more particularly to apparatuses and methods for communications channel decoding.
Mobile wireless communications systems generally require the capability of handovers between base station transceivers. However, such handovers have traditionally created technical issues resulting in dropped calls during the handover process.
One such issue with respect to GSM networks is that of maintaining control channel integrity when using low-rate Advanced Multi-rate Codec (AMR) modes, half-rate traffic channel modes, or indeed for many other logical traffic channels. For example, when operating using the full-rate AMR 5.9 bps or 4.74 kbps speech encoder modes, as well as half-rate or other modes as mentioned above, the carrier-to-interference plus noise ratio (CINR) required to maintain an acceptable frame error rate (FER) on the traffic channel (TCH) may be significantly less than the CINR required to maintain the control channel FER. Example control channels include, but are not limited to, the Slow Associated Control Channel (SACCH) and the Fast Associated Control Channel (FACCH) as specified in the GSM 3GPP specifications, among other various control channels.
Because control channel reception is critical for certain operations such as handovers, the control channel error rate is of particular importance for reducing the network dropped call rate (DCR).
It would be desirable to combine multiple control channel transmissions by for example Chase combining. However, if any bit or bits change between subsequent transmissions, the codeword resulting from forward error correction (FEC) coding methods, such as the combination of Fire coding and convolutional coding in the case of the GSM FACCH, would also change and direct combining of control channel blocks would not be possible.
One potential solution, which could be applied to the FACCH, would be to permit re-transmission of an identical message, thereby permitting the Mobile Station (MS) to combine the first and second transmissions. However, such a method presents several additional issues.
First, the method requires that the 184-bit payload (Layer 2 or “L2” message) of the first and second FACCH transmissions be exactly identical to permit Chase combining at the receiver. No modification whatsoever of the FACCH message content could be permitted.
Second, in order to permit the receiver to combine the appropriate control messages, some means of implicit or explicit signaling would need to be provided to instruct that combining should be performed. For example, the time between the first and second transmission of the FACCH frame could be an exact number, or otherwise known number, of TDMA frames.
Third, the method could only be used to combine a known limited number of FACCH transmissions. Flexible support for the Chase combining of many FACCH transmissions would prove impractical. Finally, the method could not be supported by legacy networks or provide significant advantage to legacy terminals.
Methods and apparatuses for jointly decoding messages are provided herein.
In the various embodiments, messages may be jointly decoded based on a priori known differences between initial and subsequent messages, without regard to timing or spacing between initial and subsequent messages as further described herein.
In some of the embodiments, the properties of Soft-Input, Soft-Output (SISO) decoders may be utilized for the purpose of handling and combining initial messages with subsequent messages having a priori known differences.
Further in other embodiments, the linear properties of Fire codes and convolutional codes are utilized along with a priori known differences between subsequent message transmissions such that various combining techniques may be effectively employed. In such embodiments, advantageous use is made of the linearity of the message encoding such that if a difference in subsequent message transmission is known, then a difference in codewords is also known.
If a difference between codewords is known then log likelihood ratios corresponding to the difference may be negated such that previous and subsequent channel blocks may be combined directly in the various embodiments. The various embodiments therefore enable Chase combining for signaling information that is not fully repeated.
In the various embodiments, upon receiving a first message transmission a mobile station may first attempt a general message decode without attempting to combine prior message transmissions.
If the decoding fails, the mobile station may hypothesize the message, and combine the soft decision information made available by the current and previous message frame observations consistent with the hypothesis. The soft-combining methods used by the mobile station depend on the bits that change between each hypothesized message retransmission.
If a successful message decode does not result from the first hypothesis, the mobile station may proceed to hypothesize the next message and so on up to N message transmissions. The mobile station may update the stored message soft decision information before each decode attempt.
Turning now to the drawings wherein like numerals represent like components,
The various embodiments utilize the repetition of the Layer 2 (L2) and Layer 3 (L3) content known to occur during specific message transfer sequences on the Fast Associated Control Channel (FACCH).
The illustrated signaling scenario is a typical downlink message transfer from the BTS 103 to the MS 101, in which an L3 message is carried in an L2 Information (I) frame. The precise content of the L3 message is not significant therefore the L3 message may be a handover command or any other signaling message of interest.
It is critically important to note that while the various embodiments herein disclosed are described in the context of a GSM FACCH, the embodiments are not so limited. Rather, any signaling scenario in which re-transmission is used may enjoy the benefits of the various embodiments herein disclosed. Further, the various embodiments may be applied to any wireless communications standard or air interface such as, but not limited to, GSM/GPRS/EDGE, UMTS, IEEE 802.16, IEEE 802.20, IEEE 802.11 etc.
Returning now to
In the initial I-frame 105, the P bit is set to value 0 and a send sequence number N(S), also contained within the I-frame 105, has an arbitrary value depending upon how many I frames have previously been transferred from the BTS 103 to the MS 101.
In the example illustrated by
As shown in
Further in
In
The Control field 205 further comprises a send sequence number N(S) field and a receive sequence number N(R) field. The N(S) and N(R) fields are 3 bit sequence numbers and may have any appropriate value. The Poll bit (P bit), which is bit position 5 of the control field 205, is 0 on an initial transmission of an I-frame and is set to 1 on all retransmissions of the same I frame as previously discussed above.
The length indicator field 207 indicates the size of the Layer 3 message 209. Bit position 2, of length indicator field 207, which is defined as the “More bit” (M bit) indicates whether the current L2 block is the last block or whether more L2 blocks follow and need to be concatenated to form the full L3 message. An M bit set to 1 indicates that more L2 blocks follow. The Length Extension (EL) bit, of length indication field 207, is always 1 on the FACCH. The Layer 3 message follows the length octet. Any unused octets are filled with the hexadecimal value 2B.
Turning now to
In the various embodiments, a specific sequence of L2 information messages may be hypothesized for a signaling scenario as illustrated in
Returning to
Critically important to the various embodiments, the combination of a rate-½ convolutional code and a Fire code forms a concatenated code which is linear in binary Galois Field GF(2) as illustrated in
Therefore, with respect to
c(x)=ƒ(d(x))=Md(x) (0.1)
where the function ƒ(•) is a linear operator, equivalently expressed by the matrix operator M.
Therefore an information sequence d(x) may be decomposed into:
d(x)=do(x)+Px12 (0.2)
where the polynomial do(x) has no term of order 12 (i.e. O(12)). Because ƒ(•) is a linear operator, the codeword polynomial c(x) may be expressed as:
where co(x) is the codeword excluding the P bit, referred to hereinafter as the basic codeword, and p(x) is the codeword corresponding to the P bit.
The received codeword polynomial c(x) for each possible value of P can then be expressed as:
In the various embodiments, Equation (0.4) enables combining a first and any subsequent transmissions under the signaling scenario illustrated in
More importantly, knowledge of the exact value of P may be relaxed during combining of subsequent transmissions to the assumption that either a) the value of P corresponding to LLRki for i=0 is different from the value of P corresponding to LLRki for i>0, or b) it is the same. Equation (0.4) enables the combining under either hypothesis.
Returning again briefly to the signaling scenario illustrated by
In the embodiments illustrated by
It is to be noted that equation (0.4) as described above, and with respect to the various embodiments, provides for particularly efficient combining of first and subsequent codeword observations and subsequent forward error correction decoding of the resulting combined code word observations by exploiting the properties of linear codes.
However, the present disclosure is not so limited, and it is to be understood that other methods of jointly decoding the first and subsequent codeword observations are equally applicable to the various embodiments herein disclosed. For example, a simple convolutional decoder based on the Viterbi decoding principle could be modified to operate on the first and subsequent codeword observations by embedding hypotheses on the information word differences between the first and subsequent codeword observations, such as on the P bit in the particular case of the FACCH transmission, in the branch metric computations used to construct the trellis state metrics. Therefore in some embodiments, each particular hypothesis regarding the difference in information words may be associated with the already well-known hypotheses embedded in the Viterbi decoder regarding the encoded bits delimiting each state transition in the trellis.
In the particular case where the function ƒ(•) in equation (0.1) is a non-linear code, combining methods exploiting the linear code property of equation (0.4) may not be applicable, and such alternative methods of joint decoding may be required.
It is to be noted that various other embodiments for joint decoding exist in accordance with the present disclosure. For example, in some embodiments, joint decoding may make use of a priori known differences in messages in a probabilistic fashion, noting that for “Soft-Input, Soft-Output” (SISO) decoders, a portion of the probabilistic values corresponding to the codeword bits, whether input or output to the SISO decoder, correspond to the a priori known message difference. Therefore, handling may be applied to such bit portions such that an initial and subsequent message may be jointly decoded.
It is to be understood that, although the messages are described for simplicity as “initial” and “subsequent,” such messages may not necessarily be transmitted at initial and subsequent times in all embodiments. For example, in some embodiments a set of messages with known differences in information content may be sent over a multiplicity of channels, where these channels may be time-division, frequency division or code-division multiplexed, or mapped in some other fashion onto a set of physical resources, such as Orthogonal Frequency Division Modulation (OFDM) sub-carriers. All that is required is that the receiver be aware of the differences in information word content, and the method of forward error correction encoding. In such instances, the messages may then be jointly decoded using for example, linear or SISO decoder embodiments. Further with respect to subsequent messages wherein such messages do arrive at times subsequent to an initial message, no particular time interval or spacing is required by the various embodiments.
It is to be understood that much of the inventive functionality and many of the inventive principles herein disclosed are best implemented with or in software or firmware programs or instructions and integrated circuits (ICs) such as digital signal processors (DSPs) or application specific ICs (ASICs) as is well known by those of ordinary skill in the art. Therefore, further discussion of such software, firmware and ICs, if any, will be limited to the essentials with respect to the principles and concepts used by the various embodiments.
The FACCH decoding methods herein disclosed are defined in the context of a voice call using a single timeslot per TDMA frame. Therefore, the MIPS (millions of instructions per second) and memory constraints are less restrictive than for the multi-slot case. Nonetheless, neglecting the cost of combining LLRs, the computational complexity of decoding (convolutional and Fire decoding) is doubled in some embodiments due to the need to hypothesize the first observed FACCH block which may or may not be the actual first transmission. Therefore, with respect to the memory requirements of some embodiments, two buffers of 456 words of 16 bits must be allocated to store LLR values between the reception of subsequent FACCH frames.
Therefore, turning briefly to
then C≅A+B when the first and second messages are identical and C≅Σ(∥LLR1|−|LLR2∥) when each bit in the first message is the opposite to that of the second message. In general, when both messages are different, it is reasonable to assume that half of the message bits are the same and half are different. This may be indicated when C has a value close to the midpoint 705 of the range described in
Therefore, ranges 703 and 706 of
Note that a subset Ω of message codeword LLR values over which metrics A, B, C etc. are computed may comprise any suitable subset of message codeword bits in the various embodiments. For example, Ω may comprise the entire message codeword, or only those bits in the message codeword whose value is not influenced by the value of P.
Similarly, a range of
Therefore, various implementations may be used in the various embodiments falling within the scope of the present disclosure. For example, some embodiments may only check for differences in the P-bit, while other embodiments may only check whether the overall message is different. Likewise, both the overall message and the P-bit may be checked in some embodiments.
It is to be understood that one skilled in the art may choose to apply a similar procedure utilizing a suitable threshold instead of the overall midpoint 705 or may perform a different form of algebraic computation in calculating C and the ranges in
Further, it is to be understood that while the present disclosure discusses checking the P-bit of a FACCH because the P-bit is a known difference between subsequent FACCH transmissions, the present disclosure is not so limited. Rather, the present disclosure is applicable to any situation wherein subsequent transmissions comprise a known difference in messages between an initial and the subsequent transmissions. Therefore, messages other than FACCH with known differences other than a P-bit may be decoded and combined using the various embodiments herein disclosed.
Returning therefore to
However, if the P bit has changed, then the LLR values corresponding to the P bit may be negated as shown in block 609 prior to combining in block 613.
Turning now to
Block 805 represents the determination of whether the messages are different, while block 807 represents the determination of whether a portion of the message is different. Note that for the FACCH embodiments described in detail above, this portion would correspond to the P-bit which is known to change between FACCH transmissions. However, any transmission of interest having such a known difference may employ the embodiments illustrated by
Block 809 represents handling of the probabilistic values corresponding to the codeword bit portion or portions, for a priori known message differences, as discussed above, whether input or output to the SISO decoder. One example therefore, is LLRs, wherein certain LLRs corresponding to known message differences may be negated prior to decoding.
However, important to understanding the embodiments illustrated by
The various embodiments illustrated by
Any number of bits, and/or portions of messages may be processed as illustrated by
While various embodiments have been illustrated and described, it is to be understood that the disclosure is not so limited. Numerous modifications, changes, variations, substitutions and equivalents will occur to those skilled in the art without departing from the spirit and scope as defined by the appended claims.
Number | Date | Country | Kind |
---|---|---|---|
06300059 | Jan 2006 | EP | regional |
Number | Name | Date | Kind |
---|---|---|---|
5319712 | Finkelstein et al. | Jun 1994 | A |
6067444 | Cannon et al. | May 2000 | A |
6097772 | Johnson et al. | Aug 2000 | A |
6360345 | Kim et al. | Mar 2002 | B1 |
6392572 | Shiu et al. | May 2002 | B1 |
6782035 | Nakamura et al. | Aug 2004 | B1 |
6788959 | Jokinen et al. | Sep 2004 | B2 |
7058407 | Chi et al. | Jun 2006 | B2 |
7243285 | Foisy et al. | Jul 2007 | B2 |
7260764 | Chen | Aug 2007 | B2 |
7321643 | Chang et al. | Jan 2008 | B2 |
7453907 | Jonsson et al. | Nov 2008 | B2 |
7593489 | Koshy et al. | Sep 2009 | B2 |
7600172 | Berens et al. | Oct 2009 | B2 |
7711068 | Shokrollahi et al. | May 2010 | B2 |
7813453 | Khandekar et al. | Oct 2010 | B2 |
7899051 | Bachu et al. | Mar 2011 | B2 |
8189581 | Bachu et al. | May 2012 | B2 |
20010014858 | Hirayama | Aug 2001 | A1 |
20020056064 | Kidorf et al. | May 2002 | A1 |
20020159545 | Ramesh et al. | Oct 2002 | A1 |
20020172261 | Hershey et al. | Nov 2002 | A1 |
20020186778 | Agami et al. | Dec 2002 | A1 |
20030039217 | Seo et al. | Feb 2003 | A1 |
20030072286 | Kim et al. | Apr 2003 | A1 |
20030112780 | Ouyang et al. | Jun 2003 | A1 |
20030118031 | Classon et al. | Jun 2003 | A1 |
20040179469 | Attar et al. | Sep 2004 | A1 |
20040225940 | Kerr et al. | Nov 2004 | A1 |
20050003831 | Anderson | Jan 2005 | A1 |
20050030914 | Binzel et al. | Feb 2005 | A1 |
20050030964 | Tiedemann, Jr. et al. | Feb 2005 | A1 |
20050276249 | Damnjanovic et al. | Dec 2005 | A1 |
20050276266 | Terry | Dec 2005 | A1 |
20060040706 | Wu et al. | Feb 2006 | A1 |
20060193348 | Unno et al. | Aug 2006 | A1 |
20060256792 | Kwong et al. | Nov 2006 | A1 |
20070042794 | Fischer | Feb 2007 | A1 |
20080039133 | Ma et al. | Feb 2008 | A1 |
20080072269 | Malladi et al. | Mar 2008 | A1 |
20080181163 | Ye et al. | Jul 2008 | A1 |
20080186885 | Athalye et al. | Aug 2008 | A1 |
20080294962 | Goel | Nov 2008 | A1 |
20080316995 | Bachu et al. | Dec 2008 | A1 |
20080320527 | Bachu et al. | Dec 2008 | A1 |
20090149207 | Zhang et al. | Jun 2009 | A1 |
20090168927 | Bachu et al. | Jul 2009 | A1 |
Number | Date | Country |
---|---|---|
1811674 | Jul 2007 | EP |
1811711 | Jul 2007 | EP |
2418571 | Mar 2006 | GB |
2006014365 | Jan 2006 | JP |
9825360 | Jun 1998 | WO |
9927745 | Jun 1999 | WO |
0062467 | Oct 2000 | WO |
2007087511 | Aug 2007 | WO |
2007087512 | Aug 2007 | WO |
2008000292 | Jan 2008 | WO |
Entry |
---|
Public Email Discussion Forum: 3GPP—TSG—RAN—WG1©LIST.ETSI.ORG, Discussion Group; Chmiel, M. Jun. 11, 2007 8:49 AM. |
Annamalai A et al: “Analysis of Unslotted Direct-Sequence Spread Spectrum Multiple Access Network with Packet combining” XP006007997, pp. 1673-1674. |
Miyagi M et al: “Selective Repeat Type-II Hybrid Fec/ARG Systems Using Concatenated Codes” XP000425130, pp. 25-34. |
3GPP TS 44.006 V6.2.0; Nov. 2005; 59 Pages. |
3GPP TS 44.005 V7.0.0; Nov. 2005; 32 Pages. |
Sanderovich et al., LDPC Coded MIMO Multiple Access With Iterative Joint Decoding, IEEE Transactions on Information Theory, vol. 51, No. 4, Apr. 2005, pp. 1437-1450. |
The State Intellectual Property Office of the People's Republic of China, China Patent Appln. No. 200880021002.1, Sep. 15, 2011, 10 pages. |
3GPP TS 45.003 v6.8.0 (Sep. 2005) 3rd Generation Partnership Project; Technical Specification Group GSM/EDGE, Radio Access Network, Channel coding (Release 6) 202 pages. |
Japanese Patent Office, “Office Action” (in Japanese) Japanese Pat Appln. No. 2010-51056 , Nov. 22, 2011, 3 pages. |
3GPP TSG-GERAN Meeting #25, GP-015429 Jun. 20-24, 2005. |
3GPP TS 44.0.05 V7.0.0; Nov. 2005; 32 pages. |
3GPP TS 44.0.06 V6.2.0; Nov. 2005; 59 pages. |
Motorola: “R2-061459 System Broadcast” Internet Citation May 3, 2006. |
Public Email Discussion Forum: 3GPP—TSG—RAN—WG1©LIST.ETSI.ORG, Discussion Group; Chmiel, M. Jun. 11,2007 8:49 AM. |
Public Email Discussion Forum: 3GPP—TSG—RAN—WG1©LIST.ETSI.ORG, Discussion Group; Buckley, M. Jun. 11, 2007 8:47 PM. |
Public Email Discussion Forum: 3GPP—TSG—RAN—WG1©LIST.ETSI.ORG, Discussion Group; Shiego, T. Jun. 12, 2007 10:19 PM. |
Public Email Discussion Forum: 3GPP—TSG—RAN—WG1©LIST.ETSI.ORG, Discussion Group; Jianglei, M. Jun. 13, 2007 8:35 AM. |
Public Email Discussion Forum: 3GPP—TSG—RAN—WG1©LIST.ETSI.ORG, Discussion Group; Dahlman, E. Jun. 13, 2007 9:27 AM. |
Public Email Discussion Forum: 3GPP—TSG—RAN—WG1©LIST.ETSI.ORG, Discussion Group; Jianglei, M. Jun. 14, 2007 3:53 PM. |
Public Email Discussion Forum: 3GPP—TSG—RAN—WG1©LIST.ETSI.ORG, Discussion Group; Onggosanusi, E Jun. 15, 2007 12:39 AM. |
Public Email Discussion Forum: 3GPP—TSG—RAN—WG1©LIST.ETSI.ORG, Discussion Group; Chmiel, M. Jun. 15, 2007 2:12 AM. |
Public Email Discussion Forum: 3GPP—TSG—RAN—WG1©LIST.ETSI.ORG, Discussion Group; Joonyoung, C. Jun. 18, 2007 7:07 AM. |
Annamalai A et al: “Analysis of Unslotted Direct-Sequence Spread Spectrum Multiple Access Network with Packet combining” XP006007997, pp. 1673-1674, Sep. 25, 1997 vol. 22 No. 20. |
Miyagi M et al: “Selective Repeat Type-II Hybrid Fec/ARG Systems Using Concatenated Codes” XP000425130, pp. 25-34, electronic and communications in Japan, Part I, vol. 76, No. 6, 1993. |
3GPP TS 25.212 V6.7.0, Dec. 2005; 84 pages. |
3GPP TSG RAN WG1 #49 Meeting, Kobe, Japan, May 7-11, 2007 R1-072265. |
IEEE Transactions on Information Theory, vol. IT-13, No. 2, Apr. 1967. |
United States Patent and Trademark Office, “Non-Final Office Action” for U.S. Appl. No. 11/765,909 dated Sep. 3, 2010, 13 pages. |
United States Patent and Trademark Office, “Final Office Action” for U.S. Appl. No. 11/765,909 dated Mar. 18, 2011, 14 pages. |
United States Patent and Trademark Office. “Notice of Allowance” for U.S. Appl. No. 11/967,971 dated Oct. 19, 2010, 7 pages. |
3GPP TSG-GERAN Meeting #25, GP-051429, Jun. 20-24, 2005. |
Number | Date | Country | |
---|---|---|---|
20080279312 A1 | Nov 2008 | US |