WO2017176147A1 - Device and method for adjusting transmission size in case of decoding failures - Google Patents

Abstract

The disclosure relates to a device (200) for adjusting a size for transmissions of encoded symbols over a communication link in case of decoding failures, the device (200) comprising: a Soft-Input-Soft-Output (SISO) decoder (201) configured to decode input information (L_in ^(l), 202) derived from a transmission of encoded symbols (r^(l)) over a communication link to obtain output information (L_out ^(l), 204); a failure detector (203) configured to detect a decoding failure (206) of the SISO decoder (201 ) based on an evaluation of the output information (L_out ^(l), 204); and a processor (205) configured to adjust a size (N', 208) for a next transmission of encoded symbols (r^(l+1)) based on a functional relation of an average bit-wise entropy (H) of the output information (L_out ^(l), 204) normalized with respect to the input information (L_in ^(l), 202) in case of a decoding failure (206).

Description

Device and method for adjusting transmission size in case of decoding failures

TECHNICAL FIELD

The present disclosure relates to a device and method for adjusting a size for

transmissions of encoded symbols over a communication link in case of decoding failures and to a transceiver including such a device. In particular, the invention relates to an Incremental Redundancy Adaptation (IRA) method for the HARQ (Hybrid Automatic Repeat Request) communication protocol.

BACKGROUND

The hybrid automatic repeat request (HARQ) scheme is used in a communication system to provide both efficient and reliable data transmissions. Incremental Redundancy (IR) is an HARQ method of combination of the payloads from different retransmissions. The performance of the IR is limited due to the fixed size of the retransmitted payload (for example, a fixed retransmitted payload is currently used in LTE systems as a baseline). Soft information from the output of the decoder can be used for adjusting the size of the next transmitted payload. This technique is often referred to as Reliability-Based Hybrid ARQ (RB-HARQ). State-of-art RB-HARQ algorithms suffer from some drawbacks, such as flexibility to code rate, modulation order, channel model etc., high overhead in feedback channel to indicate bit or block positions that should be retransmitted, robustness to demodulator and equalizer errors.

SUMMARY

It is the object of the invention to provide a robust and flexible scheme for reliability based communication, in particular reliability based HARQ communication.

This object is achieved by the features of the independent claims. Further implementation forms are apparent from the dependent claims, the description and the figures.

A basic idea of the invention is to apply a novel method that exploits the IR method of HARQ and uses soft information from the decoder for adaptation of the size of retransmitted payload in case of wrong decoding. Since this method is tightly related to the IR method it is named as "Incremental Redundancy Adaptation Hybrid Automatic Repeat Request (IRA HARQ)" hereinafter. This solution can be used for LTE Physical Uplink Shared Channel (PUSCH) providing superior performance.

For proper determining of the length for the next transmission a function of soft information obtained before and after the decoding process can be used. The IRA HARQ method is based on a novel algorithm which makes use of the relative entropy (also known as Kullback Leibler divergence) as a measure of the distinguishability between two distributions. One of those distributions is a soft output of the decoder that can be for example a Turbo, LDPC or convolutional code. The second one is a uniform distribution since it corresponds to the most uncertainty. Based on this calculation a size of the next payload in case of wrong decoding is provided. Additional overhead in feedback channel consists only in the number of resource blocks (RB) for next transmission (for example four information bits at most). The IRA HARQ algorithm does not rely on precalculated Look-Up-Tables and is robust with respect to the channel estimation errors. Performance is verified in downlink (DL) LTE for different channel models (EPA 5Hz , EVA 70 Hz , ETU 300Hz) and for different RB and MCS in SISO mode as presented below with respect to Figures 6 to 11.

The devices, systems and methods according to the disclosure may be applied in reliability based communication, in particular Reliability-Based Hybrid ARQ (RB-HARQ) schemes as shown in Figure 1 showing a block diagram of a communication system 100 according to an RB-HARQ scheme.

Such a communication system 100 includes a transmitter 110 and a receiver 120. The transmitter 110 transmits a sequence of transmission symbols 116, / being the transmission sequence number, over a downlink communication channel, e.g. a Rayleigh channel 150 to the receiver 120 which receives a sequence of reception symbols r 122. The transmitter 110 includes a source and cyclic redundancy check (CRC) device 111 , a channel encoder 113 and an M-QAM modulator 115. The source and cyclic redundancy check (CRC) device 111 provides a binary information sequence with attached Cyclic Redundancy Check code bits Μ₄ 112 to the channel encoder 113 which encodes this sequence 112 to provide a binary code sequence c^w 114 based on reliability information 128 indicating which bits should be retransmitted as received from the receiver 120. The M-QAM modulator 1 15 modulates the binary code sequence c^w 1 14 to provide the sequence of transmission symbols 1 16.

The receiver 120 includes an M-QAM demodulator 121 , a HARQ combiner 123 and a SISO decoder 125. The M-QAM demodulator 121 decodes the sequence of reception symbols r 122 to provide log-likelihood ratios (LLRs) 124 to the HARQ combiner

123 which provides soft input information 126 to the SISO decoder 125. In the case of correct decoding 130, e.g. checked by cyclic redundancy check (CRC) passed, the SISO decoder 125 provides output information bits 132. Otherwise in the case of a decoding failure of the SISO decoder 125, the SISO decoder 125 provides the reliability information 128 indicating which bits should be retransmitted to the transmitter 1 10, in particular to the channel encoder 1 13.

Here, "soft input information" is soft information of all bits before decoding, while "soft output information" is soft information of all bits after decoding. L_m denotes LLR values before decoding, and L_out is LLR values after decoding.

The communication system 100 depicted in Figure 1 that uses an RB-HARQ scheme can be described as follows.

Let u = ( (l), u(2), ... , (/C)) be the binary information sequence with attached Cyclic Redundancy Check code bits 1 12 of overall length K. Let N_t be the number of code bits at Z-th transmission. So after channel encoding 1 13 at transmission / we have a binary code sequence c⁽') = (c^l(l), c^l(2), ... , c^l(Ni 1 14. Map this sequence 1 14 to M-QAM modulated sequence x^w 1 16. After passing through fully interleaved Rayleigh channel 150 let r⁽ⁱ⁾ be the vector of received complex symbols 122:

where h * is Rayleigh fading channel coefficient with zero mean and unit variance and n ^l) is the complex Gaussian noise with variance 2 - af.

The M-QAM demodulator 121 calculates channel log-likelihood ratios (LLRs) 124 and can be implemented in Max-Log MAP fashion, so max0._ei4:0 _] σ_χ, 0_;·) ,

where

where and a_t ² are estimations of fading coefficient and noise variance respectively, A- constellation points of M-QAM, k = 1 ... , log₂ M

After that HARQ combining 123 is followed where input LLRs 124 summed at code positions that was previously sent (Chase combining) and LLRs for new parity bits just concatenated to form one codeword (Incremental Redundancy), and that codeword goes to the Soft Input Soft Output (SISO) channel decoder 125. Examples of this decoder 125 are Turbo, LDPC or convolutional code decoders. So corresponds to input LLRs 126 of SISO decoder 125 at /-th transmission and L®_t " ^soft output LLRs of the decoder 125. All RB-HARQ algorithms take L^_ut and in case of decoding failure (CRC fails) try to determine which bits should be retransmitted 128 and signal it in the feedback channel.

The devices, systems and methods according to the disclosure may include HARQ combiner for processing a HARQ communication protocol.

Hybrid automatic repeat request (HARQ) is a combination of high-rate forward error- correcting coding and ARQ error-control. In standard ARQ, redundant bits are added to data to be transmitted using an error-detecting (ED) code such as a cyclic redundancy check (CRC). Receivers detecting a corrupted message request a new message from the sender. In Hybrid ARQ, the original data is encoded with a forward error correction (FEC) code, and the parity bits are either immediately sent along with the message or only transmitted upon request when a receiver detects an erroneous message.

The devices, systems and methods according to the disclosure may include soft decision decoders and SISO decoders for decoding Soft input information to Soft output information, in particular Soft input information received from a HARQ combiner. A soft-decision decoder is using a class of algorithms to decode data that has been encoded with an error correcting code. Whereas a hard-decision decoder operates on data that take on a fixed set of possible values (typically 0 or 1 in a binary code), the inputs to a soft-decision decoder may take on a whole range of values in-between. This extra information indicates the reliability of each input data point, and is used to form better estimates of the original data. Therefore, a soft-decision decoder will typically perform better in the presence of corrupted data than its hard-decision counterpart.

A soft-input soft-output (SISO) decoder is a type of soft-decision decoder used with error correcting codes. "Soft-input" or "soft input information" refers to the fact that the incoming information or data may take on or may include values other than 0 or 1 , in order to indicate reliability. "Soft-output" refers to the fact that each bit in the decoded output also takes on or includes a value indicating reliability. The soft output may be used for example as the soft input to an outer decoder in a system using concatenated codes, or to modify the input to a further decoding iteration such as in the decoding of turbo codes. Examples include the BCJR algorithm and the soft output Viterbi algorithm.

In particular, in the context of the present invention, the expression "soft information of bit U" does not refer to the value of U (i.e., U=0 or U=1), but rather to probabilities (P(U=0) and P(U=1) = 1 - P(U=0)). The "LLR of bit U" is defined as LLR(U) = log (P(U=0) / P(U=1 )). When referring to reliability, the following is meant: bit U is more reliable than bit V, if |LLR(U)| >= |LLR(V)|. This amounts to making a hard decision (i.e., a hard decision: if P(U=0) > P(U=1), then U=0, otherwise U = 1). With other words, soft information of a bit is equivalent to reliability of a bit (both indicating probabilities), and soft value of a bit is equivalent to LLR of a bit (indicating some real values, which can be calculated as LLR(U) = log (P(U=0) / P(U=1)).

An example may illustrate above aspects: given two bits U=0 and V=1. Then after BPSK modulation U is mapped into U - 1 , and V into V = -1. Suppose after AWGN channel (U" = U'+ N(0,1), V" = V'+ N(0,1)), U" = 1.2 and V" = -0.7. Based on these values probabilities are calculated as follows: P(U=0|U") = 1 - P(U = 1 |U"). For simplicity the notation of conditional probability is omitted, resulting in P(U=0) = 1 - P(U = 1)). Finally, LLR(U) and LLR(V) van be calculated. One can check that the hard decision of U is more reliable than that the hard decision of V. In order to describe the invention in detail, the following terms, abbreviations and notations will be used:

BS: Base Station, eNodeB

UE: User Equipment, e.g. a mobile device or a machine type communication device

IRA: Incremental Redundancy Adaptation

HARQ: Hybrid Automatic Repeat Request

RB: Reliability Based

LTE: Long Term Evolution

PUSCH: Physical Uplink Shared Channel

CRC: Cyclic Redundancy Check

ACK: Acknowledgement

NACK: Non- Acknowledgement

RF: Radio Frequency

SISO: Soft Input Soft Output

MAP: Maximum A Posteriori

According to a first aspect, the invention relates to a device for determining a size for transmissions of encoded symbols over a communication link in case of decoding failures, the device comprising: a Soft-Input-Soft-Output decoder configured to decode input information derived from a transmission of encoded symbols over a communication link to obtain output information; a failure detector configured to detect a decoding failure of the Soft-Input-Soft-Output (SISO) decoder based on an evaluation of the output information; and a processor configured to adjust a size for a next transmission of encoded symbols based on a functional relation of an average bit-wise entropy of the output information normalized with respect to the input information in case of a decoding failure.

The size for the next transmission adjusted by the processor can be a number of data bits or additional data bits that are required to be transmitted for the next transmission.

Alternatively, the size can be a number of transmit data symbols or additional transmit data symbols that are required to be transmitted for the next transmission. When the additional bits or symbols are delivered, the receiver can provide these additional bits or symbols in another redundancy version, e.g. according to the HARQ protocol. Ν^', i.e. the size of the next transmission, is calculated at the receiver, then this size (just a number) is transmitted over a control feedback channel to the transmitter, then the transmitter reads out from a circular buffer N' code bits with another starting position (i.e., redundancy version).

Redundancy versions that are used for retransmissions can be set in advance. In the context of LTE standard the redundancy version refers to the starting position in the circular buffer, such that different parity bits can be transmitted on the different

transmission. At the receiver the parity bits from different redundancy versions, that is, soft information about the transmitted bits, are combined into one codeword.

Such a device provides a robust and flexible scheme for reliability based communication, in particular reliability based HARQ communication.

In a first possible implementation form of the device according to the first aspect, the next transmission is a retransmission according to a Hybrid Automatic Repeat Request (HARQ) scheme, in particular according to one of the following HARQ schemes: an Incremental Redundancy (IR) HARQ scheme, a HARQ type III scheme, a HARQ scheme according to Long Term Evolution (LTE). This provides the advantage that the device provides robust and flexible adaptation of multiple transmissions for all kinds of HARQ algorithms, such as IR HARQ, HARQ type III, LTE HARQ and many other HARQ schemes.

In a second possible implementation form of the device according to the first aspect as such or according to the first implementation form of the first aspect, the size for the next transmission is variable, in particular different from a size of at least one previous transmission.

This provides the advantage that the device provides flexible adaptation of the next transmission; for each transmission an independently computed optimal size can be determined.

In a third possible implementation form of the device according to the first aspect as such or according to any one of the preceding implementation forms of the first aspect, the processor is configured to adjust the size for the next transmission based on an adaptive algorithm.

This provides the advantage that the processor can adaptively provide the optimal transmission size, thereby saving complexity.

In a fourth possible implementation form of the device according to the first aspect as such or according to any one of the preceding implementation forms of the first aspect, the failure detector is configured to detect the decoding failure based on a cyclic redundancy check (CRC) of the output information.

This provides the advantage that the decoding failure can be easily and efficiently detected when using a standard device such as a CRC circuit. In a fifth possible implementation form of the device according to the first aspect as such or according to any one of the preceding implementation forms of the first aspect, both, the input information and the output information comprise values indicating a reliability of the respective information. Reliability values can be real values but also quantified values. This provides the advantage that the decoder can process soft information based on reliability of the processed information. Such a decoder and hence the device has a higher accuracy than a decoder or device based on deterministic information. Through this application, soft information refers to log likilihood ratio (LLRs). Soft input information refers to LLRs prior decoding obtained from the demodulator. Soft output information refers to LLRs obtained after decoding.

In a sixth possible implementation form of the device according to the first aspect as such or according to any one of the preceding implementation forms of the first aspect, the processor is configured to adjust the size for the next transmission based on a function of the average bit-wise entropy and a total length of previous transmissions.

This provides the advantage that the device can consider previous transmission and thus increase the accuracy of decoding. In a seventh possible implementation form of the device according to the sixth

implementation form of the first aspect, the processor is configured to adjust the size N' of a next transmission by computing a new value of N' as N' = F(f(\ -H),N) , where N is l -x

the sum of lengths of all previous transmissions, and F(x,y) = y , and

x

( ) = -0.6x² + 1.45x .

The size N' is adjusted based on the function F and function f, that is, N' =

F(f( ^~H),N) _unctj_0n f j_{s a} smooth function dependent on H . Preferably, the function f is quadratic with respect to H .

In the case where (1 - H) is very close to 0 (or H is near 1) and the decoded codeword is not confirmed, by using the function f to adjust the size of the next transmission, it is guaranteed that still an amount of information, sufficient for ensuring an error free decoding with high probability, is sent in the next payload. Using the quadratic function f increases the probability of correct decoding after retransmission, in case in a first transmission the decoded codeword is not confirmed by the CRC.

In an eighth possible implementation form of the device according to the first aspect as such or according to any one of the preceding implementation forms of the first aspect, the symbols are encoded based on an LTE turbo code.

This provides the advantage that the device can be flexibly applied for LTE

communications, i.e. in common communication networks. In a ninth possible implementation form of the device according to the first aspect as such or according to any one of the preceding implementation forms of the first aspect, the Soft-Input Soft-Output decoder comprises a Max-Log MAP decoder.

If a soft demodulator provides error in estimation of LLRs such that the LLRs are multiplied by some constant factor, then the Max-Log MAP decoder will be robust to such kind of errors. For such errors the method will provide the same result as well as in the case we don't get a multiplication error after the soft demodulation. For example, if an SNR estimated error is obtained from the demodulator, since the proposed method mitigates this effect the error will not be propagated to the Max-Log MAP decoder.

According to a second aspect, the invention relates to a transceiver, comprising:

a receiver, configured to receive a transmission of encoded symbols over a

communication link; a demodulator, in particular an M-QAM demodulator, configured to provide a plurality of log-likelihood ratios (LLRs) based on a demodulation of the transmission of encoded symbols; a hybrid automatic repeat request (HARQ) combiner configured to provide input information based on the plurality of log-likelihood ratios; and a device according to the first aspect as such or according to any one of the preceding implementation forms of the first aspect, configured to adjust a size for a next

transmission of encoded symbols in case of a decoding failure.

Such a transceiver provides a robust and flexible scheme for reliability based

communication, in particular reliability based HARQ communication.

In a first possible implementation form of the transceiver according to the second aspect, the transceiver comprises a transmitter configured to provide the size for the next transmission over a feedback channel to request a transmission of a size given by a number of additional information in another redundancy version.

This provides the advantage that the size for the next transmission can be efficiently signaled to the transmitter, e.g. a transmitter of the base station, by using the feedback channel independently from the uplink data/control channel. When using the feedback channel, the uplink channel for providing uplink data to the transmitter can be unchanged, i.e. no feedback information has to be included in the uplink data symbols.

In a second possible implementation form of the transceiver according to the second aspect as such or according to the first implementation form of the second aspect, the transmitter is configured to provide the size for the next transmission over an LTE

Physical Uplink Shared Channel (PUSCH).

This provides the advantage that the size for the next transmission can be efficiently signaled over the PUSCH channel which is a channel available in LTE for signaling control information in uplink direction. According to a third aspect, the invention relates to a method for indicating a size for a retransmission of encoded symbols over a communication link in case of decoding failures, the method comprising: receiving a transmission of encoded symbols over a communication link; deriving input information from the transmission of encoded symbols; decoding the input information based on Soft-Input-Soft-Output (SISO) decoding to obtain output information; detecting a decoding failure by evaluating the output information; and indicating a size for a next transmission of encoded symbols based on a functional relation of an average bit-wise entropy of the output information normalized with respect to the input information in case of a decoding failure.

Such a method provides a robust and flexible scheme for reliability based communication, in particular reliability based HARQ communication. The method provides superior spectral efficiency gain.

According to a fourth aspect, the invention relates to a method of handling a

retransmission of a hybrid automatic repeat request (HARQ) for a receiver in a

communication system, the method comprising: receiving a block of data of error correcting code (ECC) over a communication link; determining a resource index (Rl) to indicate a size for a payload in the next reception by performing decoding to obtain soft output information (output LLR's) using Soft-Input-Soft-Output decoder, then normalizing soft output information obtained, then calculating the mean value H of the entropy of normalized soft output information and then computing a function F of H and of sum of resource indices of all previous transmissions (Rl⁺); requesting a quantity F(H, Rl⁺) of additional information associated with the data block.

This method provides a robust and flexible scheme for reliability based communication, in particular reliability based HARQ communication. The spectral efficiency gain is superior to the baseline HARQ algorithm.

In a first possible implementation form of the method according to the fourth aspect, the step of normalizing output soft information is comprised of dividing all output LLR's by the mean of absolute values of input LLR's. In a second possible implementation form of the method according to the fourth aspect as such or according to the first implementation form of the fourth aspect, the step of determining Rl for the next transmission makes use of the following functions:

H(x) = -x^■ log₂ x - (1 - x)^■ log₂(l - x) ; l ^{i = 1}°⁸² P u(i) = l)'

where the summation takes place over all systematic bits and K is the number of such bits;

R T * _Rm D I+ _ r n_T+ _. (1+0-6C1-H)²- 1.45(1-H))

RI_next = F(H, RI ) - [ RI ₍__{0 6(1}__{H)2+ li45(1}__H) I In a third possible implementation form of the method according to the fourth aspect as such or according to any of the preceding implementation forms of the fourth aspect, the requested additional information bits are read from Circular Buffer Rate Matching (CBRM) with another redundancy version (RV).

In a fourth possible implementation form of the method according to the fourth aspect as such or according to any of the preceding implementation forms of the fourth aspect, the ECC is a LTE turbo code.

In a fifth possible implementation form of the method according to the fourth aspect as such or according to any of the preceding implementation forms of the fourth aspect, the Soft Input Soft Output decoder is a Max-Log-Map decoder.

The disclosed methods avoid most of the problems described above. The methods may use soft output information from the decoder to determine the size of the next transmitted payload in case of wrong decoding. In simple words, a size of the next transmitted payload is the number of parity bits that should be generated in case of the Incremental Redundancy protocol. The disclosed algorithm is tested in a communication system with physical layer similar to LTE Physical Downlink Shared Channel which uses a

combination of Chase combining and IR scheme (so called HARQ type III) and Turbo- coding. For the downlink case scenario additional overhead may be required in Uplink Control Channel to signal the number of Resource Blocks for next transmission. In one implementation form this overhead may be estimated as 4bits. For Uplink Shared

Channel, base station may calculate the number of Resource Blocks (RBs) for user retransmission and reschedule these RBs for user in Downlink Control Channel. That means that the disclosed algorithm can be used for LTE Physical Uplink Share Channel. The spectral efficiency gain is about 6-8% which can be achieved by applying the disclosed method independent from channel condition, modulation order, coding rate and channel estimation errors with small complexity at the receiver.

BRIEF DESCRIPTION OF THE DRAWINGS

Further embodiments of the invention will be described with respect to the following figures, in which: Fig. 1 shows a block diagram illustrating a communication system 100 according to a Reliability-Based Hybrid ARQ (RB-HARQ) scheme;

Fig. 2 shows a block diagram illustrating a device 200 for adjusting a size for

transmissions of encoded symbols over a communication link according to an

implementation form;

Fig. 3 shows a block diagram illustrating a transceiver 300 according to an implementation form; Fig. 4 shows a block diagram illustrating a receiver path 400 according to an

implementation form;

Fig. 5 shows a schematic diagram illustrating a method 500 for adjusting a size for a retransmission of encoded symbols over a communication link in case of decoding failures according to an implementation form;

Fig. 6 shows a performance diagram 600 illustrating spectral efficiency of an Incremental Redundancy Adaptation Hybrid Automatic Repeat Request (IRA HARQ) algorithm according to the disclosure with parameters RB 20, MCS 14 and channel "EPA 5Hz"; Fig. 7 shows a performance diagram 700 illustrating spectral efficiency of an IRA HARQ algorithm according to the disclosure with parameters RB 20, MCS 14 and channel "EVA 70Hz"; Fig. 8 shows a performance diagram 800 illustrating spectral efficiency of an IRA HARQ algorithm according to the disclosure with parameters RB 20, MCS 14 and channel "ETU 300Hz";

Fig. 9 shows a performance diagram 900 illustrating spectral efficiency of an IRA HARQ algorithm according to the disclosure with parameters RB 10, MCS 5-6-7-8-9 and channel "EVA 70Hz";

Fig. 10 shows a performance diagram 1000 illustrating spectral efficiency of an IRA HARQ algorithm according to the disclosure with parameters RB 25, MCS 21-22-23-24-25 and channel "ETU 300Hz"; and

Fig. 11 shows a performance diagram 1100 illustrating spectral efficiency of an IRA HARQ algorithm according to the disclosure with parameters RB 10, MCS 11-12-13-14-15 and channel "EVA 70Hz".

DETAILED DESCRIPTION OF EMBODIMENTS

In the following detailed description, reference is made to the accompanying drawings, which form a part thereof, and in which is shown by way of illustration specific aspects in which the disclosure may be practiced. It is understood that other aspects may be utilized and structural or logical changes may be made without departing from the scope of the present disclosure. The following detailed description, therefore, is not to be taken in a limiting sense, and the scope of the present disclosure is defined by the appended claims. It is understood that comments made in connection with a described method may also hold true for a corresponding device or system configured to perform the method and vice versa. For example, if a specific method step is described, a corresponding device may include a unit to perform the described method step, even if such unit is not explicitly described or illustrated in the figures. Further, it is understood that the features of the various exemplary aspects described herein may be combined with each other, unless specifically noted otherwise.

Fig. 2 shows a block diagram illustrating a device 200 for determining a size for transmissions of encoded symbols over a communication link in case of decoding failures according to an implementation form.

The device 200 includes a Soft-Input-Soft-Output (SISO) decoder 201 , a failure detector 203 and a processor 205. The Soft-Input-Soft-Output (SISO) decoder (201) is configured to decode input information L_in ^(l), 202 derived from a transmission of encoded symbols r^(l) over a communication link, e.g. a Rayleigh channel 150 as described with respect to Fig. 1 , to obtain output information L_out ^(l), 204. The failure detector 203 is configured to detect a decoding failure 206 of the SISO decoder 201 based on an evaluation of the output information L_out ^(l), 204. The processor 205 is configured to adjust a size N', 208 for a next transmission of encoded symbols r^(l+1) based on a functional relation of an average bitwise entropy H of the output information L_out^(l), 204 normalized with respect to the input information L_in ^(l), 202 in case of a decoding failure 206.

The next transmission may be a retransmission according to a Hybrid Automatic Repeat Request (HARQ) scheme, for example, according to an Incremental Redundancy (IR) HARQ scheme, a HARQ type III scheme or a HARQ scheme according to Long Term Evolution (LTE). The size ', 208 for the next transmission may be variable, for example different from a size of a previous transmission. The processor 205 may adjust the size N', 208 for the next transmission based on an adaptive algorithm, e.g. an adaptive algorithm as described below.

The failure detector 203 may detect the decoding failure 206 based on a cyclic

redundancy check (CRC) of the output information L_out^(l), 204. The failure detector 203 may include a CRC circuit for detecting a decoding failure 206 of the SISO decoder 201.

Both, the input information L_in ^(l), 202 and the output information L_out ^(l), 204 may include soft information, i.e. including or taking on values indicating a reliability of the respective information. The processor 205 may adjust the size N', 208 for the next transmission based on a function F(x,y) of the average bit-wise entropy H and a total length N of previous transmissions, e.g. as described below. The function F(x,y) of the average bit-wise entropy Ή and the total length N may be quadratic with respect to the average bit-wise entropy H , e.g. as described below. The processor 205 may adjust the function F(x,y) of the average bit-wise entropy H and the total length N based on the relation:

F(f(l - H),N) with f(x) = -0.6x² + 1.45x , e.g. as described below.

The symbols may be encoded based on an LTE turbo code. The Soft-Input Soft-Output decoder 201 may include a Max-Log MAP decoder.

The disclosed novel algorithm may include the following steps:

Step 1 : Obtaining soft input information from the HARQ combining module after /- transmission. Define the average value

i= l

where N is length of soft input information L®.

Step 2: Performing decoding to obtain soft output information (output LLR's L _t ⁼

Step 3: If the decoded codeword is not confirmed, e.g., it fails the CRC check, following the next step.

Step 4: Normalizing soft output information by dividing all output LLR's V by the mean value ΐ

= 1(2) f(A ) . In the context of the algorithm it is important to carry out normalization, since otherwise the algorithm will not be robust in terms of signal-to-noise ratio (SNR). Observe that doing that allows to stabilize the mean value of absolute values of LLR, but the variation of LLR remains dependent of variance of noise. In fact, the obtained LLR like values do not correspond to the actual a-posteriori probabilities computed by the decoder.

Step 5: Calculating the average bit-wise entropy H of normalized soft output information

where the Shannon entropy is defined as usual

H(x) A -x^■ log₂ x - (1 - x)^■ log₂(l - x) and the probability values P(u(i) = 0) can be obtained via the following equations

'^(ί) - ^log2

^{1,2 K}- Notice that the average bit-wise relative entropy (denoted by D(P\ \Q)) between normalized soft output information and uniform distribution (P(u(i) = 0) = P(u(i) = 1) = can be directly expressed as a linear function of the avera e bit-wise entropy. Indeed:

In case Q (Q = -, then D(P\ \Q) = 1 - H(P).

Such an approach is motivated next. In case that after decoding the codeword is not confirmed and there is a distribution which is close to the uniform distribution, it is evident that in a given channel condition it will be practically impossible to transmit and correctly decode the information with a chosen coding rate. One can see that such a case will be reflected at the value Ή. More precisely, it will be close to 1 , and this implies that a relatively big amount of information on the next payload is needed. On the other hand, if the value Ή is close to 0 and the codeword is not confirmed, then a relatively small amount of information may be sent on the next payload. It may be noted that using entropy is not a unique way for this problem. For example, for a fixed channel conditions and code rate the above approach looks equivalent to measuring an LLR gain after decoding (ratio between average absolute value of input LLR's and output LLR's). But this value lies between zero and infinity, and there is no universal tool for calculating a size for the next transmitted payload.

Step 6: Computing a value N' = F(f(l - H), N), where N is the sum of lengths of all previous transmissions (it may be the sum of resource indices or something else, for example) and

F(x, y) = y ^■

f(x) = -0.6x² + 1.45x. The motivation for doing that is the following. It may be assumed that the value (1 - H) (or more precisely (1 - f(H))) stands for "a relative share of obtained information" (the full information is K, while only (1 - H)K is obtained). So, if the condition of the channel will remain the same, then the size N' of the next portion of additional redundancy bits should be equal to the root of the equation:

(It is known that for I R HARQ scheme the accumulated mutual information SNRL SNR₂) for two transmissions, where SNR for the first channel use is denoted by SNR^ for the second - SNR₂, and a channel has capacity C(SNR), describes by l(SNRi, SNR₂) = CiSNR,) + C(SNR₂).)

The function f(x) provides robustness to the solution. The function was chosen in such a way that it is a quadratic function and f(0) = 0, f(l) = 0.85, f(0.75) = 0.75. Suppose that (1 - H) is very close to 0 (or H is near 1) and the decoded codeword is not confirmed, then without the function f(x) an insufficient amount of information may be sent in the next payload. This situation is extremely rare, but when it occurs an algorithm has to work in a correct way.

Step 7: Requesting the quantity N' of additional information read from Circular Buffer Rate Matching (CBRM) with another redundancy version (RV). It was said that 4 bits are sufficient to signal over the feedback channel. The set P of possible lengths for the next payload can be set as follows:

p (N 2N 16 Π

l 8 ' 8 8 J '

then ceil N' to the nearest element of P, and request the corresponding quantity of additional bits. In case N' > 2N, the quantity 2N of additional bits may be requested.

Fig. 3 shows a block diagram illustrating a transceiver 300 according to an implementation form. The transceiver 300 includes a demodulator 303, in particular an M-QAM

demodulator, e.g. an M-QAM demodulator 121 as described above with respect to Fig. 1 , a hybrid automatic repeat request (HARQ) combiner 305, e.g. a HARQ combiner 123 as described above with respect to Fig. 1 , and a device 200, e.g. as described above with respect to Fig. 1 , for adjusting a size N', 208 for a next transmission of encoded symbols r^(l+1) in case of a decoding failure 206. The receiver 301 is configured to receive a transmission of encoded symbols η^(|), 302 over a communication link, e.g. a channel 150 as described above with respect to Fig. 1. The demodulator 303 is configured to provide a plurality of log-likelihood ratios Lj^(l), 304 based on a demodulation of the transmission of encoded symbols η^(|), 302. The HARQ combiner 305 is configured to provide input information L_in ^(l), 202 based on the plurality of log- likelihood ratios 304. The device 200 is configured to adjust a size NT, 208 for a next transmission of encoded symbols r^(l+1) in case of a decoding failure 206.

The transceiver 300 may include a transmitter for providing the size N', 208 for the next transmission over a feedback channel 407 to request a transmission of a size N', 208 number of additional information in another redundancy version. The transmitter may provide the size N' 208 for the next transmission over an LTE Physical Uplink Shared Channel (PUSCH).

Fig. 4 shows a block diagram illustrating a receiver path 400 according to an

implementation form. The receiver path 400 includes a Soft-Input Soft-Output (SISO) decoder, e.g. a SISO decoder 201 as described above with respect to Fig. 2, a processor for determining the length N', e.g. a processor 205 as described above with respect to Fig. 2, a CRC circuit, e.g. a failure detector 202 as described above with respect to Fig. 2, and a feedback channel 407. The Soft-Input-Soft-Output (SISO) decoder 201 is configured to decode soft input information L_in, 202 derived from a transmission of encoded symbols over a

communication link, e.g. a Rayleigh channel 150 as described with respect to Fig. 1 , to obtain output information L_out, 204.

The soft input information L_in, 202 may include reliabilities of code bits from the output of demodulator, e.g. the M-QAM demodulator 121 described above with respect to Fig. 1. The CRC circuit 202 describes the block for detecting the decoding error.

The soft input information L_out, 204 may include reliabilities of systematic bits after decoding, i.e. after the decoder block 201. The messages ACK/NACK describe fulfillment of the CRC condition or CRC condition is failed. N/N' may indicate the length of sent bits / next transmission (e.g., RB number, resource index). The CRC circuit 203 is configured to detect a decoding failure of the SISO decoder 201 by comparing a CRC (cyclic redundancy check) of the output information L_out, 204 with a CRC field included in the soft input information L_in, 202 or in the transmission of encoded symbols. If both CRC values are the same, the CRC check is passed and an ACK message 402 is sent via the feedback channel 407 to the transmitter, e.g. the transmitter 110 as described above with respect to Fig. 1. If both CRC values are different, the CRC check is not passed and a NACK message 404 is sent to the processor 205 which is configured to adjust a size N', 208 for a next transmission of encoded symbols based on a functional relation of an average bit-wise entropy H of the soft output information L_out, 204 normalized with respect to the soft input information L_in, 202 in case of a decoding failure, i.e. a NACK 404 received from the CRC circuit 203. The processor 205 is configured to send the NACK message 404 together with the size ', 208 for the next transmission via the feedback channel 407 to the transmitter, e.g. the transmitter 110 as described above with respect to Fig. 1. The next transmission may be a retransmission according to a Hybrid Automatic Repeat Request (HARQ) scheme, for example, according to an Incremental Redundancy (IR) HARQ scheme, a HARQ type III scheme or a HARQ scheme according to LTE. The size N', 208 for the next transmission may be variable, for example different from a size of a previous transmission. The processor 205 may adjust the size 1ST, 208 for the next transmission based on an adaptive algorithm, e.g. an adaptive algorithm as described above with respect to Fig. 2.

Both, the soft input information L_in, 202 and the soft output information L_out ^(l), 204 may include soft information, i.e. including or taking on values indicating a reliability of the respective information.

The processor 205 may adjust the size N', 208 for the next transmission based on a function F(x,y) of the average bit-wise entropy H and a total length N of previous transmissions, e.g. as described above with respect to Fig. 2. The function F(x,y) of the average bit-wise entropy H and the total length N may be quadratic with respect to the average bit-wise entropy H , e.g. as described above with respect to Fig. 2. The processor 205 may adjust the function F(x,y) of the average bit-wise entropy H and the total length N based on the relation: (/(1 -H),N) with f{x) = -0.6x² + 1.45x , e.g. as described above with respect to Fig. 2.

The symbols may be encoded based on an LTE turbo code. The Soft-Input Soft-Output decoder 201 may include a Max-Log MAP decoder. The receiver path 400 may be implemented for LTE like systems with a small overhead in the feedback channel 407.

Fig. 5 shows a schematic diagram illustrating a method 500 for indicating a size for a retransmission of encoded symbols over a communication link in case of decoding failures according to an implementation form. The method includes: receiving 501 a transmission of encoded symbols, e.g. a transmission η^(ι), 302 as described above with respect to Fig. 3, over a communication link, e.g. a channel 150 as described above with respect to Fig. 1 . The method 500 further includes deriving 502 input information L_in ^(l), e.g. input information Li^(l), 304 as described above with respect to Fig. 3, from the transmission of encoded symbols π ⁾, 302.

The method 500 further includes decoding 503 the input information Li^(l) based on Soft- Input Soft-Output (SISO) decoding to obtain output information, e.g. output information L_out ^(l), 204 as described above with respect to Fig. 2. The method 500 further includes detecting 504 a decoding failure by evaluating the output information, e.g. by using a failure detector 203 as described above with respect to Fig. 2 or a CRC circuit 203 as described above with respect to Fig. 4. The method 500 further includes adjusting 505 a size for a next transmission of encoded symbols based on a functional relation of an average bit-wise entropy H of the output information normalized with respect to the input information in case of a decoding failure.

The disclosed method 500 allows to increase overall spectral efficiency for a cell. The additional overhead in the feedback channel for downlink transmission can be reduced to only 4 bits, for uplink it can be applied without additional overhead. One can see that the algorithm is not dependent of a channel model, a modulation scheme, a coding rate. The spectral efficiency can be further improved if the step of determining the length of the next payload will be dependent of a coding rate and a modulation scheme. With other words the spectral efficiency is improved when the modulation and coding rate of the

transmission is taken into account. The method 500 is verified in different channel conditions as shown below with respect to Figures 6 to 11.

For the simulations performed with respect to Figures 6 to 11 , the following restrictions apply: An LTE turbo code with 8 iterations and scaled Max Log MAP decoding (scale factor = 0.75 for all iterations) is used. CRC of length 24 is used. The maximal number of transmissions is limited to 2 (in case of wrong decoding after the first transmission, the second one is used). Two decoding algorithms are compared with each other: The baseline algorithm, which uses the same length for the second payload (if needed) and is denoted as "baseline HARQ", and the disclosed algorithm as described above with respect to Fig. 2, which is denoted as "IRA HARQ" in the figures. It should be mentioned that the block error rate after two transmissions is less than 10^~4 in the presented simulations. The Signal to noise ratio (SNR) in all graphics measured on symbol after CRC is attached and will be referred to as SNR (dB). The y-axis illustrates the spectral efficiency.

Fig. 6 shows a performance diagram 600 illustrating spectral efficiency of an Incremental Redundancy Adaptation Hybrid Automatic Repeat Request (IRA HARQ) algorithm 601 according to the disclosure with parameters RB 20, MCS 14 and channel "EPA 5Hz" in comparison to the baseline HARQ algorithm 602. The following parameters are applied: modulation coding scheme ( CS) index is 14, modulation is 16-QAM, data size K is 5160, coding rate (CR^ for the first transmission is 0.4696, channel bandwidth (CBw) is 5MHz, the number of resource blocks (RB) is 14, the LTE channel model used is "EPA 5Hz", type of equalizer used is "Minimum mean square error" (MMSE), ideal channel estimation (ICE), and SNR region is [3.5, 7]. Figure 6 shows the superior spectral efficiency of the IRA HARQ algorithm 601 over the baseline HARQ algorithm 602.

Fig. 7 shows a performance diagram 700 illustrating spectral efficiency of an IRA HARQ algorithm 701 according to the disclosure with parameters RB 20, MCS 14 and channel "EVA 70Hz" in comparison to the baseline HARQ algorithm 702. The following

parameters are applied: MCS = 14, modulation is 16-QAM, K = 5160, CRi = 0.4696, CBw = 5MHz, RB = 14, the LTE channel model used is "EVA 70Hz", MMSE equalizer, ICE estimation, and SNR region is [4.5, 7]. Figure 7 shows the superior spectral efficiency of the IRA HARQ algorithm 701 over the baseline HARQ algorithm 702.

Fig. 8 shows a performance diagram 800 illustrating spectral efficiency of an IRA HARQ algorithm according to the disclosure with parameters RB 20, MCS 14 and channel "ETU 300Hz" in comparison to the baseline HARQ algorithm 702. The following parameters are applied: MCS = 14, modulation is 16-QAM, K = 5160, CRi = 0.4696, CBw = 5MHz, RB = 14, the LTE channel model used is "ETU 300Hz", MMSE equalizer, ICE estimation, and SNR region is [3.5, 5.5]. Figure 8 shows the superior spectral efficiency of the IRA HARQ algorithm 801 over the baseline HARQ algorithm 802. Fig. 9 shows a performance diagram 900 illustrating spectral efficiency of an IRA HARQ algorithm 901 according to the disclosure with parameters RB 10, MCS 5-6-7-8-9 and channel "EVA 70Hz" in comparison to the baseline HARQ algorithm 902. The following parameters are applied: The parameter RB = 10 is fixed, the channel used is "EVA 70Hz", MMSE equalizer, ICE estimation and modulation is QPSK. Five adjacent modulation coding schemes {5,6,7,8,9} are tested (this means that the corresponding data sizes are {872, 1032, 1224, 1384, 1554}, and coding rates for the first transmissions are {0.3246, 0.3826, 0.4522, 0.5101 , 0.5715} ). SNR region is [-4.6, 0.4]. For each SNR point such a modulation coding scheme is applied, which maximizes spectral efficiency. Figure 9 shows the superior spectral efficiency of the IRA HARQ algorithm 901 over the baseline HARQ algorithm 902. The IRA HARQ algorithm 901 provides about 5-8% spectral efficiency gain or 0.3-0.5dB in Power.

Fig. 10 shows a performance diagram 1000 illustrating spectral efficiency of an IRA HARQ algorithm 1001 according to the disclosure with parameters RB 25, MCS 21-22-23-24-25 and channel "ETU 300Hz" in comparison to the baseline HARQ algorithm 1002. The following parameters are applied: The following parameters are fixed: RB = 25, CBw = 5MHz, the LTE channel model used is "ETU 300Hz", MMSE equalizer, ICE estimation and modulation is 64-QAM. Five adjacent modulation coding schemes are tested: {21, 22, 23, 24, 25} (this means that the corresponding data sizes are {10680, 11448, 12576, 13536, 14112}, and coding rates for the first transmissions are {0.5171 , 0.5542, 0.6087, 0.6551 , 0.6829}). SNR region is [12, 17]. For each SNR point such a modulation coding scheme is applied, which maximizes spectral efficiency. Figure 10 shows the superior spectral efficiency of the IRA HARQ algorithm 1001 over the baseline HARQ algorithm 1002. The IRA HARQ algorithm 1001 provides about 3-6% spectral efficiency gain or 0.3- 0.8dB in Power.

Fig. 11 shows a performance diagram 1100 illustrating spectral efficiency of an IRA HARQ algorithm 1101 according to the disclosure with parameters RB 10, MCS 11-12-13-14-15 and channel "EVA 70Hz" in comparison to the baseline HARQ algorithm 1102. The following parameters are applied: The following parameters are fixed: RB = 10, CBw = 5MHz, the LTE channel model used is "EVA 70Hz", MMSE equalizer, real channel estimation and modulation is 16-QAM. Five adjacent modulation coding schemes are tested: {11 , 12, 13, 14, 15} (this means that the corresponding data sizes are {1554, 1736, 2024, 2280, 2536}, and coding rates for the first transmissions are {0.2859, 0.3188,

0.3710, 0.4174, 0.4638}). SNR region is [3.5, 7]. For each SNR point such a modulation coding scheme is applied, which maximizes spectral efficiency. Figure 11 shows the superior spectral efficiency of the IRA HARQ algorithm 1101 over the baseline HARQ algorithm 1102. The IRA HARQ algorithm 1101 provides about 6-8% spectral efficiency gain or 0.5-0.7dB in Power.

The present disclosure also supports a computer program product including computer executable code or computer executable instructions that, when executed, causes at least one computer to execute the performing and computing steps described herein, in particular the steps of the method 500 described above with respect to Figure 5. Such a computer program product may include a readable non-transitory storage medium storing program code thereon for use by a computer. The program code may perform the method 500 described above with respect to Fig. 5. While a particular feature or aspect of the disclosure may have been disclosed with respect to only one of several implementations, such feature or aspect may be combined with one or more other features or aspects of the other implementations as may be desired and advantageous for any given or particular application. Furthermore, to the extent that the terms "include", "have", "with", or other variants thereof are used in either the detailed description or the claims, such terms are intended to be inclusive in a manner similar to the term "comprise". Also, the terms "exemplary", "for example" and "e.g." are merely meant as an example, rather than the best or optimal. The terms "coupled" and "connected", along with derivatives may have been used. It should be understood that these terms may have been used to indicate that two elements cooperate or interact with each other regardless whether they are in direct physical or electrical contact, or they are not in direct contact with each other.

Although specific aspects have been illustrated and described herein, it will be

appreciated by those of ordinary skill in the art that a variety of alternate and/or equivalent implementations may be substituted for the specific aspects shown and described without departing from the scope of the present disclosure. This application is intended to cover any adaptations or variations of the specific aspects discussed herein.

Although the elements in the following claims are recited in a particular sequence with corresponding labeling, unless the claim recitations otherwise imply a particular sequence for implementing some or all of those elements, those elements are not necessarily intended to be limited to being implemented in that particular sequence.

Many alternatives, modifications, and variations will be apparent to those skilled in the art in light of the above teachings. Of course, those skilled in the art readily recognize that there are numerous applications of the invention beyond those described herein. While the present invention has been described with reference to one or more particular embodiments, those skilled in the art recognize that many changes may be made thereto without departing from the scope of the present invention. It is therefore to be understood that within the scope of the appended claims and their equivalents, the invention may be practiced otherwise than as specifically described herein.

Claims

CLAIMS:

1. A device (200) for adjusting a size for transmissions of encoded symbols over a communication link in case of decoding failures, the device (200) comprising: a Soft-Input-Soft-Output (SISO) decoder (201 ) configured to decode input information (L_in ^(l), 202) derived from a transmission of encoded symbols (r^(l)) over a communication link to obtain output information (L₀ut^(l), 204); a failure detector (203) configured to detect a decoding failure (206) of the SISO decoder (201 ) based on an evaluation of the output information (L_ou,^(l), 204); and a processor (205) configured to adjust a size (Ν', 208) for a next transmission of encoded symbols (r^{(l+ )}) based on a functional relation of an average bit-wise entropy ( H ) of the output information (L_out^(l), 204) normalized with respect to the input information (L_in ^(l), 202) in case of a decoding failure (206).

2. The device (200) of claim 1 , wherein the next transmission is a retransmission according to a Hybrid Automatic

Repeat Request (HARQ) scheme, in particular according to one of the following HARQ schemes: an Incremental Redundancy (IR) HARQ scheme, a HARQ type III scheme, a HARQ scheme according to Long Term Evolution (LTE).

3. The device (200) of claim 1 or 2, wherein the size (Ν', 208) for the next transmission is variable, in particular different from a size of at least one previous transmission.

4. The device (200) of one of the preceding claims, wherein the processor (205) is configured to adjust the size (NT, 208) for the next transmission based on an adaptive algorithm.

5. The device (200) of one of the preceding claims, wherein the failure detector (203) is configured to detect the decoding failure (206) based on a cyclic redundancy check (CRC) of the output information (L_out ^(l), 204).

6. The device (200) of one of the preceding claims, wherein both, the input information (L_in ^(l), 202) and the output information (L₀^ 204) comprise values indicating a reliability of the respective information.

7. The device (200) of one of the preceding claims, wherein the processor (205) is configured to adjust the size (Ν', 208) for the next transmission based on the functional relationship given by a function (F(H ,N)) of the average bit-wise entropy (H ) and a total length (N) of previous transmissions.

8. The device (200) of claim 7, wherein the processor (205) is configured to adjust the size NT by computing a new value of N' as N' = F(f(l - H),N) , where N is the sum of lengths of all previous

1 - x

transmissions, and F(x,y) = y , and /(x) = -0.6x² + 1.45 .

x

9. The device (200) of one of the preceding claims, wherein the symbols are encoded based on an LTE turbo code.

10. The device (200) of one of the preceding claims, wherein the Soft-Input Soft-Output decoder (201 ) comprises a Max-Log MAP decoder.

1 1 . A transceiver (300), comprising: a receiver (301 ), configured to receive a transmission of encoded symbols (η^(|), 302) over a communication link; a demodulator (303), in particular an M-QAM demodulator, configured to provide a plurality of log-likelihood ratios (L , 304) based on a demodulation of the transmission of encoded symbols (η^(|), 302); a hybrid automatic repeat request (HARQ) combiner (305) configured to provide input information (L_in ^(l), 202) based on the plurality of log-likelihood ratios (Li^(l), 304); and a device (200) according to one of claims 1 to 1 1 , configured to adjust a size (NT, 208) for a next transmission of encoded symbols (r^(l+1)) in case of a decoding failure (206).

12. The transceiver (300) of claim 1 1 , comprising: a transmitter configured to provide the size (Ν', 208) for the next transmission over a feedback channel (407) to request a transmission of a size (NT, 208), number of additional information in another redundancy version.

13. The transceiver of claim 1 1 or 12, wherein the transmitter is configured to provide the size (Ν', 208) for the next transmission over an LTE Physical Uplink Shared Channel (PUSCH).

14. A method (500) for indicating a size for a retransmission of encoded symbols over a communication link in case of decoding failures, the method comprising: receiving (501 ) a transmission of encoded symbols (r^(l)) over a communication link; deriving (502) input information (L_in ^(l)) from the transmission of encoded symbols

(r^(,)); decoding (503) the input information (L_in ^(l)) based on Soft-Input-Soft-Output (SISO) decoding to obtain output information (L_out ^(l)); detecting (504) a decoding failure by evaluating the output information (L_out^(l)); and adjusting (505) a size for a next transmission of encoded symbols (r^(l+1)) based on a functional relation of an average bit-wise entropy ( H ) of the output information (L₀ut^(l)) normalized with respect to the input information (L_in ^(l)) in case of a decoding failure.