WO2020030115A1

WO2020030115A1 - Downlink transmission with two-stage feedback: early prediction-based feedback of physical downlink shared channel and hybrid automatic repeat request feedback

Info

Publication number: WO2020030115A1
Application number: PCT/CN2019/099997
Authority: WO
Inventors: Trung Kien Le; Umer Salim
Original assignee: JRD Communication (Shenzhen) Ltd.
Priority date: 2018-08-10
Filing date: 2019-08-09
Publication date: 2020-02-13
Also published as: GB201813092D0; GB2576210B8; GB2576210B; CN112292825A; GB2576210A; GB2576210A8

Abstract

A method of data transmission between a base station and a user equipment. The method comprises, for the base station: receiving, at the base station, data to be transmitted to the user equipment; transmitting the data to the user equipment; receiving a signal indicative of early prediction feedback from the user equipment. Based on the signal indicative of early prediction feedback, if the signal indicative of early prediction indicates DTX, retransmitting the data to the user equipment repeating the method from the step of receiving a signal indicative of early prediction feedback. If the signal indicative of early prediction indicates NAK, determining, whether a sufficient latency budget is available decided. If the budget is available, continuing the method at the step of receiving an HARQ feedback. If the budget is not available, retransmitting the data to the user equipment and repeating the method from the step of receiving a signal indicative of early prediction feedback. The method further comprises receiving an HARQ feedback and, if the HARQ feedback indicates NAK, retransmitting the data to the user equipment and repeating the method from the step of receiving a signal indicative of early prediction feedback. The method comprises, for the user equipment: receiving data from the base station; estimating an error probability based on at least a portion of the received data; generating a signal indicative of early feedback based on the estimated error probability; and sending the signal indicative of early prediction feedback to the base station. A base station and a user equipment configured to perform the respective methods are also disclosed.

Description

Downlink transmission with two-stage feedback: early prediction-based feedback of Physical Downlink Shared Channel and Hybrid Automatic Repeat Request feedback

Technical Field

The following disclosure relates to downlink transmission with two-stage feedback: early prediction-based feedback of Physical Downlink Shared Channel (PDSCH) and Hybrid Automatic Repeat Request (HARQ) . In particular, the following disclosure relates to a transmission scheme that combines both the early feedback based on prediction and the conventional Hybrid Automatic Repeat Request (HARQ) feedback in order to exploit the advantages as well as mitigate the disadvantages of both techniques.

Background

Wireless communication systems, such as the third-generation (3G) of mobile telephone standards and technology are well known. Such 3G standards and technology have been developed by the Third Generation Partnership Project (3GPP) . The 3rd generation of wireless communications has generally been developed to support macro-cell mobile phone communications. Communication systems and networks have developed towards a broadband and mobile system.

Figure 1 shows a schematic diagram of an example of three base stations forming a cellular network. In cellular wireless communication systems User Equipment (UE) is connected by a wireless link to a Radio Access Network (RAN) . The RAN comprises a set of base stations which provide wireless links to the UEs located in cells covered by the base station, and an interface to a Core Network (CN) which provides overall network control. As will be appreciated the RAN and CN each conduct respective functions in relation to the overall network. For convenience the term cellular network will be used to refer to the combined RAN &CN, and it will be understood that the term is used to refer to the respective system for performing the disclosed function.

The 3rd Generation Partnership Project has developed the so-called Long Term Evolution (LTE) system, namely, an Evolved Universal Mobile Telecommunication System Territorial Radio Access Network, (E-UTRAN) , for a mobile access network where one or more macro-cells are supported by a base station known as an eNodeB or eNB (evolved NodeB) . More recently, LTE is evolving further towards the so-called 5G or NR (new radio) systems where one or more cells are supported by a base station known as a next generation NodeB (gNB) . NR is proposed to utilise an Orthogonal Frequency Division Multiplexed (OFDM) physical transmission format.

A trend in wireless communications is towards the provision of lower latency and higher reliability services. For example, NR is intended to support Ultra-reliable and low-latency communications (URLLC) and massive Machine-Type Communications (mMTC) are intended to provide low latency and high reliability for small packet sizes (typically 32 bytes) . A user-plane latency of 1ms has been proposed with a reliability of 99.99999%, and at the physical layer a packet loss rate of 10 ^-5 or 10 ^-6 has been proposed.

mMTC services are intended to support a large number of devices over a long life-time with highly energy efficient communication channels, where transmission of data to and from each device occurs sporadically and infrequently. For example, a cell may be expected to support many thousands of devices.

This disclosure relates to the downlink (DL) transmission scheme of control and data for URLLC of the 5G NR system. URLLC has the strict requirements of reliability and latency that are much tighter than those of the legacy 4G system, LTE. It leads to a challenge in design of control and data channels on the grounds that there typically is a trade-off between reliability and latency. In order to achieve low latency, a short packet is transmitted but causes a degradation in channel coding and a decrease in reliability. On the other hand, to attain reliability, more resources are consumed when a longer packet with more parity check bits is transmitted, potentially with multiple retransmissions. Due to the constraint of latency, the number of retransmissions is limited and might be not sufficient to achieve the specified reliability. This means that a mechanism helping the gNB make the decision to retransmit the packets faster is necessary so that the retransmission opportunity increases.

The emergence of new paradigms like connected self-driving cars, automated industrial control, as well as augmented and virtual reality led the wireless standardization bodies to take these into account. To that respect, 3GPP has defined three service paradigms for 5G. Those services include Enhanced Mobile Broadband (eMBB) for high data rate transmission, URLLC for devices requiring low latency and high link reliability and mMTC to support a large number of low-power devices for a long life-time requiring highly energy-efficient communication. These services have diverse requirements on latency, reliability, massive connection density, and energy efficiency and have resulted in the introduction of many new techniques in the Rel-15 of 5G NR.

For URLLC service requirements, the term reliability is defined in 3GPP TR 38.802 as follows: “Reliability can be evaluated by the success probability of transmitting X bytes within a certain delay, which is the time it takes to deliver a small data packet from the radio protocol layer 2/3 SDU ingress point to the radio protocol layer 2/3 SDU egress point of the radio interface, at a certain channel quality (e.g., coverage-edge) . ”

The latency in physical layer can be expressed by the sum of 4 terms:

T _L = T _ttt + T _prop + T _proc + T _reTx (1)

where:

· T _ttt: time-to-transmit latency, is the time required to transmit a packet,

· T _prop: propagation latency, is the time for a signal to travel from the transmitter to the receiver,

· T _proc: processing time for channel estimation, encoding, decoding of the first transmission, and

· T _reTx: time for the retransmission.

It is also noted in 3GPP TR 38.802 that spectral efficiency and energy consumption also should be considered when trying to achieve a reliability target.

The reliability requirement for URLLC is specified in 3GPP TR 38.913: “Ageneral URLLC reliability requirement for one transmission of a packet is 10 ^-5 for 32 bytes with a user plane latency of 1ms. ”

In one-shot transmission, the probability of a successful transmission is calculated by the following equation:

where

· P is the successful probability of a transmission,

·

is the error probability of the physical downlink control channel (PDCCH) , and

·

is the error probability of the PDSCH.

From this equation, the error probability of a transmission is:

According to the URLLC requirement, the error probability is smaller than 10 ^-5. From equation (3) , in order to attain that probability, it is required that the error probabilities of PDCCH and PDSCH are below 10 ^-6. The design of control and data channels to achieve this value is sophisticated and consumes much time and frequency resource so as to increase reliability of the channels.

The complexity of channel design can be relaxed by considering a scheme with two transmissions. The successful probability of a transmission is calculated in R1-1701595 from ZTE:

P = P _cP _d1+ (1-P _c) P _DTXP _cP _d1+P _c (1-P _d1) P _NP _cP _d2 (4)

where:

· The first term is the probability that the initial transmission is successfully received by UE. P _c is the probability of a successful PDCCH transmission and P _d1 is the probability of successful transmission of single data transmission without any HARQ combining at the receiver.

· The second term is the probability that the second transmission is successfully received by the UE, given the UE fails to detect the PDCCH for the initial transmission. Here, P _DTX = Prob {DTX or NAK is detected | UE does not send ACK/NACK} (DTX indicating a discontinuous transmission; NAK indicating a negative acknowledgement) .

· And the third term is the probability that the second transmission is successfully received by the UE, given the UE successfully detects the PDCCH for the initial transmission but fails to decode PDSCH for the initial transmission. Here, P _N = Prob {DTX or NACK is detected | UE sends NAK} and P _d2 is the probability of successful decoding of PDSCH with the re-transmission.

From equation (4) , the reliability of URLLC can be achieved by one of the combinations of channel reliabilities as calculated in R1-1701595:

· Combination 1: P _c = P _DTX = P _N = P _d1 = P _d2 = 0.999.

· Combination 2: P _c = 0.9999, P _N = P _d1 = P _d2 = 0.999, P _DTX = 0.99.

· Combination 3: P _c = P _N = P _d2 = 0.9999, P _DTX = P _d1 = 0.99.

The requirements for control and data channel design are relaxed remarkably in comparison to a one-shot transmission. However, these values are still higher than the LTE probability of 0.99 and new techniques are necessary to enhance the performance of the channels so that the URLLC reliability target can be attained with a minimum number of retransmissions due to the time constraint.

It has been noted above that the information associated with URLLC users has extremely stringent latency and reliability requirements. Currently, there are three directions that can be used to improve the reliability in NR:

· MIMO: spatial diversity, beamforming, interference cancellation.

· Lowering the code rate of PDCCH:

· Using a higher aggregation level for PDCCH transmission to increase parity check bits.

· Providing a new downlink control information (DCI) format, which follows a compact format to reduce the number of information bits.

· Multiple transmissions.

For the hardware-related aspects, such as providing the transmitter and the receiver with a larger number of antennas, there are practical limits. The same holds true for allocating more power to the control channels, since a network cannot go beyond the regulatory limits.

Rel-15 of NR targeted the following strategies to improve the reliability of PDCCH to meet the URLLC requirements:

To ensure the reliability requirement of NR-PDCCH for URLLC, at least the following aspects should be supported:

· Defining a compact DCI format targeting a low block error rate (BLER) operation,

· The highest aggregation level should target a BLER of Y for this compact DCI format

· FFS Y, Y<1%

· FFS highest aggregation levels, e.g., 16, 32

· FFS other enhancements.

Figure 2 shows a diagram 200 illustrating PDCCH Performance for Higher ALs and Compact downlink control information (DCI) in accordance with prior art examples, and Figure 3 shows examples of URLLC operation with PDCCH. The

graphs

201, 202, 203, 204, 205, 206, 207, 208, 209, and 210 indicate the respective BLER performance. In NR, the ALs of PDCCH can reach 16 or 32. This guarantees the reliability of PDCCH by improving BLER performance by about 1 to 2 dB but also consumes much time and frequency resources in a control resource set (CORSET) . If a DCI is sent by using AL 32, it needs 32 × 6 = 192 physical resource blocks (PRBs) over 1 symbol. For a sub-carrier spacing Δf of 30 kHz, 1 PRB occupies a bandwidth of 30 kHz ×12 = 360 kHz. If the system operates with the bandwidth 40 MHz, it has around 100 PRBs per symbol. In case CORESET occupies 2 symbols, there are 200 PRBs in total while the required PRBs for 1 DCI with AL 32 are already 192 PRBs. It also has to notice that only a part of PRBs in a CORESET is the PDCCH candidates in the search space so the available PRBs for PDCCH are even less than 200 PRBs. For this reason, there is no resource for other UEs or even the considered UE. Therefore, if a single PDCCH transmission already consumes all of the resources in a CORESET, it will block the transmissions of PDCCHs of other URLLC UEs. This means that these UEs have to wait until the next transmission occasion with an available CORESET to transmit and the requirement of low latency might not be satisfied.

Above, the latency and reliability requirements for URLLC users, the techniques to achieve these requirements, and the associated trade-offs have been discussed. What should be noted, in particular, from what has been discussed above, is that repetition-based transmissions are necessary for URLLC. The gNB decides to retransmit the packet based on HARQ feedback. However, in the HARQ process, the HARQ round trip time (RTT) , which is the time interval between receiving the initial transmission and the retransmission, is large because of transmission time and processing time. For this reason, HARQ RTT may be a bottleneck for DL URLLC performance when the latency requirement is small (e.g. only 1 ms for URLLC services) . A large HARQ RTT may prevent the gNB from making a sufficient number of the necessary retransmissions within the latency budget, resulting in the inability to satisfy the target requirement. Figure 4 shows a blind repetition scheme for comparison.

Summary

This Summary is provided to introduce a selection of concepts in a simplified form that are further described below in the Detailed Description. This Summary is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used as an aid in determining the scope of the claimed subject matter.

In a first aspect there is provided a method of data transmission between a base station and a user equipment. The method comprises receiving, at the base station, data to be transmitted to the user equipment; transmitting the data to the user equipment; receiving a signal indicative of early prediction feedback from the user equipment. Based on the signal indicative of early prediction feedback: if the signal indicative of early prediction indicates a discontinuous transmission, DTX, the method further comprises retransmitting the data to the user equipment repeating the method from the step of receiving a signal indicative of early prediction feedback. If the signal indicative of early prediction indicates a negative acknowledgement, NAK, the method further comprises determining, whether a sufficient latency budget is available. If the budget is available, the method further comprises continuing the method at the step of receiving a hybrid automatic repeat request, HARQ, feedback. If the budget is not available, the method further comprises retransmitting the data to the user equipment and repeating the method from the step of receiving a signal indicative of early prediction feedback. The method further comprises receiving an HARQ feedback, and, if the HARQ feedback indicates NAK, retransmitting the data to the user equipment and repeating the method from the step of receiving a signal indicative of early prediction feedback.

In a second aspect according to aspect 1, the step of transmitting the data or of retransmitting the data to the user equipment comprises transmitting the data using a physical downlink control channel, PDCCH, and a physical downlink shared channel, PDSCH.

In a third aspect according to any one of the preceding aspects, the step of receiving an HARQ feedback further comprises, if the HARQ feedback indicates ACK and a step of retransmitting the data has been triggered, stopping retransmitting the data.

In a fourth aspect according to any one of the preceding aspects, the step of receiving an HARQ feedback comprises waiting for the HARQ feedback, if a sufficient latency budget is available.

In a fifth aspect according to any one of the preceding aspects, the step of determining, whether a sufficient latency budget is available is based on ultra-reliable and low-latency communications, URLLC, requirements. Optionally, the URLLC requirements include a reliability measure, a latency measure, a channel quality measure, a measure of available resources; and/or a traffic measure.

In a sixth aspect, there is provided a method of data transmission between a base station and a user equipment. The method comprises receiving, at the user equipment, data from the base station; estimating an error probability based on at least a portion of the received data; generating a signal indicative of early feedback based on the estimated error probability; and sending the signal indicative of early prediction feedback to the base station.

In a seventh aspect according to aspect 6, the step of estimating an error probability is triggered as soon as at least a portion of the data on PDSCH has been received.

In an eight aspect according to any one of aspects 6 or 7, the method comprises decoding the data received from the base station. Optionally, the estimated error probability is indicative of an error probability occurring in the step of decoding the data received from the base station.

In a ninth aspect according to any one of aspects 6 to 8, estimating the error probability comprises:

determining a log likelihood ratio, LLR, for each bit in a codeword of the data as:

where

L _k is the LLR of the k ^th bit in the codeword, b _k is the decoded bit of the k ^th bit in the codeword, and r _k is the received signal of the k ^th bit in the codeword.

In a tenth aspect according to any one of aspects 8 or 9, the steps of estimating the error probability and of decoding the data are performed substantially in parallel.

In an eleventh aspect according to any one of aspects 6 to 10, the error probability of a decoded bit is determined as:

where

L’ _k is the LLR of the k ^th bit.

In a twelfth aspect according to the preceding aspect, estimating the error probability is further based on a block error rate, BLER, as:

where

M is a length of a codeword of the data used to predict the error probability of the step of decoding the data received from the base station.

In a thirteenth aspect according to the preceding aspect, estimating the error probability based on the BLER _estimate, further includes setting a threshold th and generating the early prediction feedback indicating ACK or NAK as:

In a fourteenth aspect according to the preceding aspect and aspect 8, the method further comprises adapting the threshold in order to control a ratio of false positives and false negatives. Optionally, the false positives are indicative of a situation in which an early prediction indicates correct decoding of a codeword while the step of decoding fails and/or the false negatives are indicative of a situation in which an early prediction indicates incorrect decoding of a codeword while the step of decoding succeeds.

In a fifteenth aspect according to the preceding aspect, adapting the threshold is based on system level requirements; the threshold is determined by the user equipment; and/or the threshold is determined based on a command issued by the base station.

In a sixteenth aspect according to any one of aspects 6 to 15, the step of estimating an error probability is further based on a signal received earlier as decoded downlink control information, DCI.

In a seventeenth aspect, there is provided a base station configured to perform the method according to any one of aspects 1 to 5.

In an eighteenth aspect there is provided a user equipment configured to perform the method according to any one of aspects 6 to 15.

In another aspect, there is provided a non-transitory computer-readable medium. The non-transitory computer readable medium may comprise at least one from a group consisting of: a hard disk, a CD-ROM, an optical storage device, a magnetic storage device, a Read Only Memory, a Programmable Read Only Memory, an Erasable Programmable Read Only Memory, EPROM, an Electrically Erasable Programmable Read Only Memory and a Flash memory.

Brief description of the drawings

Further details, aspects and embodiments of the invention will be described, by way of example only, with reference to the drawings. Elements in the figures are illustrated for simplicity and clarity and have not necessarily been drawn to scale. Like reference numerals have been included in the respective drawings to ease understanding.

Figure 1 shows a schematic diagram of three base stations forming a cellular network;

Figure 2 shows PDCCH Performance for Higher ALs and Compact DCI;

Figure 3 shows URLLC operation with PDCCH;

Figure 4 shows a blind repetition scheme;

Figure 5 shows the impact of T _FB on the latency of regular HARQ feedback and the amount of time saved based on prediction-based feedback;

Figure 6 shows a time analysis of prediction-based feedback and regular HARQ feedback;

Figure 7 shows how missing early feedback triggers an early transmission;

Figure 8 shows a downlink transmission with early feedback and the gNB sensing latency budget;

Figure 9 shows how the gNB stops a retransmission triggered by early NAK after receiving ACK HARQ feedback;

Figure 10 shows a flowchart of the downlink transmission process with early prediction feedback;

Figure 11 shows a prediction error rate; and

Figure 12 shows error rate’s false positives.

Detailed description of the preferred embodiments

Those skilled in the art will recognise and appreciate that the specifics of the examples described are merely illustrative of some embodiments and that the teachings set forth herein are applicable in a variety of alternative settings.

Figure 1 shows a schematic diagram of three base stations (for example, eNB or gNBs depending on the particular cellular standard and terminology) forming a cellular network. Typically, each of the base stations will be deployed by one cellular network operator to provide geographic coverage for UEs in the area. The base stations form a Radio Area Network (RAN) . Each base station provides wireless coverage for UEs in its area or cell. The base stations are interconnected via the X2 interface and are connected to the core network via the S1 interface. As will be appreciated only basic details are shown for the purposes of exemplifying the key features of a cellular network.

The base stations each comprise hardware and software to implement the RAN’s functionality, including communications with the core network and other base stations, carriage of control and data signals between the core network and UEs, and maintaining wireless communications with UEs associated with each base station. The core network comprises hardware and software to implement the network functionality, such as overall network management and control, and routing of calls and data.

The HARQ process allows the gNB to carry out the retransmissions to increase the overall reliability of the transmission in the specified time constraint when the codewords in the initial transmission and the retransmissions are combined to generate a new codeword with better SNR and/or a lower code rate. However, the number of retransmissions is limited because of HARQ latency and latency requirement of URLLC. Therefore, HARQ latency must be reduced to create more retransmission occasions.

HARQ latency consists of:

· τ: propagation delay,

· T _TTI: transmission time interval duration,

· T _FB: generating feedback time that includes the decoding time for the whole received signal,

· T _A/N: transmission time for ACK/NAK, and

· T _Tx: processing time of the feedback at the gNB.

In these terms, T _FB can be reduced to lower HARQ latency. The other terms are fixed due to the inherent characteristics or hard to reduce. τ depends on the distance between the gNB and the UE and cannot be changed. Similarly, T _A/N and T _TTI are also fixed. T _Tx is hard to improve when normally feedback contains only 1 bit so the decoder at the gNB is already able to decode HARQ feedback quickly.

For comparison, Figure 5 shows the impact of T _FB on the latency of regular HARQ feedback and the amount of time saved based on prediction-based feedback in accordance with embodiments of the present invention. As can be seen in figure 5, T _FB has heavy contributions from the transmission time during which the transmitter is transmitting the packet and the processing time at the receiver during which the receiver is doing the receive processing. The receive processing includes but is not limited to equalization, demodulation, and channel decoding where channel decoding at the decoder is quite an onerous task. It impedes the opportunity of the useful retransmissions in the time constraint.

In an effort to optimize the HARQ feedback timing and to create more (re-) transmission opportunities within a certain latency target, it is proposed to use a two-stage feedback. The first stage of feedback is based on a prediction about the success/failure of decoding data on PDSCH. This early prediction-based feedback is designed so that it can be transmitted by the receiver very quickly, possibly even before the complete reception of PDSCH. It informs the transmitter about the result of the channel decoder by an intelligent estimation without passing through the entire decoding process. The predictor evaluates the error probability based on LLR estimation by using a fraction of the transmitted transport block instead of using the message passing algorithm for the whole codeword as in the decoder. It reduces computation complexity of the predictor and makes decision-making time decrease significantly compared to the full decoding process.

Figure 6 shows a time analysis of prediction-based feedback and regular HARQ feedback in accordance with embodiments of the present invention. In order to reduce the impact of the TTI duration in the HARQ RTT, the scheme proposes that the receiver only uses a fraction of the transport block signal to predict the outcome. This means that the receiver does not need to wait until the end of TTI for the complete reception of the transport block to start processing (i.e. decoding or predicting) the codeword, rather it can start the prediction computation just after receiving a fraction of the transport block, way before the complete reception of the transport block. As can be seen in figure 6, data is transmitted in TTI containing 4 OFDM symbols. In HARQ process, the time from the arrival of data in the UE to the generation of feedback is calculated by:

T = T _TTI + T _FB (5)

On the other hand, in early the feedback process, the time from the beginning of the data transmission in the gNB to the generation of feedback in the UE is calculated by:

T’= r _p × T _TTI + T _predictor (6)

where r _p: the ratio between the code used for prediction and the whole transmitted code

Equations (5) and (6) show that T’ is much smaller than T when only a fraction of T _TTI is counted in feedback generating time. Another reason is that T _predictor is also smaller than T _FB when the computation of the predictor is less complex than that of the decoder. Thus, early feedback can be generated before the UE receives the whole transmitted signal from the gNB and the retransmission is likely to start much earlier than the scheme with regular HARQ feedback.

In a first proposal, the UE starts to estimate the error probability of the decoding process based on LLR estimation just after receiving a fraction of data on PDSCH and generates early feedback to the gNB. The gNB can use early feedback to trigger an immediate retransmission.

To further improve the reliability of the prediction, the UE can exploit the parameters deduced from the early decoded signal. For example, after successfully decoding the DCI on the PDCCH, the UE has the information on channel quality and the block error rate of a codeword given transmitting power and noise. It can use this information in supplementing the prediction of decoding the PDSCH.

In a second proposal, the UE can use other metrics from earlier signal as decoded DCI to support the reliability of the prediction.

The number of symbols saved by early prediction are calculated in the following analysis. Data is assumed to be transmitted in 4 OFDM symbols with SCS 60kHz. The maximum processing time of the decoder is 0.125ms that is equivalent to 7 OFDM symbols with SCS 60kHz (1 slot with SCS 60kHz has 14 OFDM symbols spread in 0.25ms) . The predictor only uses half of the transmitted bits in order to estimate the error probability of the decoder so it only uses 2 symbols and time equivalent to 2 symbols is saved. Moreover, the predictor only deals with a small portion of bits and runs few iterations of message passing algorithm in comparison to a whole codeword and full iterations of the decoder. Thereby, the predictor takes less time to provide an estimation. If the iterations of the predictor are one fifth that of the decoder, the processing time of the predictor is 0.025ms being approximate to 2 symbols. This means that 5 symbols are saved. In total, time corresponding to 7 symbols is saved and this amount can be translated to one more retransmission in future if necessary. If the UE continues to fail to decode data several times, the accumulated saving time after each NAK feedback even leads to many retransmission opportunities in the time constraint.

Figure 7 shows how missing early feedback triggers an early transmission in accordance with embodiments of the present invention. Early prediction feedback can be very helpful to indicate the gNB not only about potential PDSCH failure but also PDCCH failure. As shown in figure 7, the gNB transmits PDCCH and PDSCH in the downlink (C1 and D1) but the UE fails to decode PDCCH so it does not know the location of PDSCH. Thereby, in this case, as it does not even know if it has been scheduled, it will send neither early prediction-based feedback nor the 2 ^nd stage of the classical feedback. As there is no prediction transmitted to the gNB and the gNB detects this discontinuous signal. Subsequently, the gNB retransmits immediately PDCCH and PDSCH (C2 and D2) instead of waiting the UE’s HARQ feedback (F2) . The prediction even can be transmitted before the end of URLLC transmission so using this strategy helps the system save much time and generates more retransmission occasions. Thus, this early prediction-based feedback will make the transmission much more robust against the PDCCH errors as an absence of prediction feedback at the gNB will be a direct indication of missed PDCCH. Although absence of prediction-based feedback may be the result of the following two factors: (1) UE missed the PDCCH, (2) gNB missed the early prediction-based feedback. As here the transmissions are for URLLC applications with stringent latency and reliability target, we will propose that the gNB makes a quick retransmission in case of not receiving early feedback. In an alternative method, to rule out the gNB missing the early feedback, if there is enough latency budget, the gNB may decide to wait for the 2 ^nd stage classical feedback before making a retransmission.

In a third proposal, early prediction feedback is used as an indicator of a success or a failure in decoding PDCCH. In case an early feedback is missing, the gNB detects a DTX and starts a retransmission immediately instead of waiting for a normal time interval of HARQ feedback.

Early feedback is generated from the following process. When the PDSCH comes to the UE, after receiving a part of the codeword (apart of the TTI) , the UE starts to predict the error probability of the decoding process. The rest of the codeword is still received by the UE in parallel to serve as the input of the decoder. Subsequently, the UE calculates the LLR for each bit in the predictor as:

where

L _k: LLR of the k ^th bit in the codeword

b _k: the decoded bit of the k ^th bit in the codeword

r _k: the received signal of the k ^th bit in the codeword.

Using the base graph as defined in 3GPP standard, the UE knows the connection among the bits used in the predictor so it runs the message passing algorithm for a small number of iterations to make the codeword converge and enhances the accuracy of the prediction. The error probability of a decoded bit is calculated by:

where

L’ _k: LLR of the k ^th bit after some message passing iterations.

After calculating the error probabilities of decoded bits, the UE estimates block error rate (BLER) by:

where

M: the length of codeword used in the predictor.

From BLER _estimate, the UE can predict the error probability of the decoder by setting a threshold (th) and generates early ACK/NAK as follows:

The prediction allows the receiver to transmit a very fast response to the transmitter regarding the success/failure of the transport block, even before receiving the full data of the transport block. This helps the gNB have more chances to retransmit the packet to boost the reliability. However, the predictor also makes the system suffer from false prediction. There are two kinds of false prediction: false negative (FN) and false positive (FP) . False negative occurs when an early NAK is sent while the decoder decodes correctly the codeword. It causes a waste of resources due to the unnecessary retransmissions but it does not affect the reliability directly. False positive occurs when an early prediction ACK is sent but, in fact, the decoder fails to decode data. This means that there is no retransmission and the packet is lost. It affects the performance of URLLC transmission. Therefore, false positive is more severe than false negative. The probabilities of false negative and false positive can be adapted by changing the proper threshold in predicting ACK or NAK. The ratio between false positive and false negative is changed depending on requirements and tolerance of the system. The threshold is adapted based on the specification and the status of the channel, available resource and energy. If there is an abundance of resource and energy and reliability is prioritized, the system is able to accept the unnecessary retransmissions due to false negative so the threshold is adapted to make false negative happen more but false positive happen less and vice versa when the threshold is adapted to make false negative happen less but false more positive happen more if resource needs to be shared among many UEs with strict latency and reliability requirements.

In a fourth proposal, threshold in feedback prediction is adapted to control the probabilities of false negative and false positive following the requirements at system level.

In order to avoid the harmful effects of false prediction and take advantage of early feedback’s benefits, a scheme combining both early feedback and regular HARQ feedback is proposed. The gNB is able to use HARQ feedback and then switches to early feedback when a fast retransmission is required to achieve the reliability requirement. The moment to switch from single stage classic HARQ feedback to two-stage feedback consisting of the first stage of early prediction feedback and the second stage of classic feedback is decided by the gNB. When two-stage feedback is activated, the gNB has to inform the users about the resources for both stages of feedback. To keep the things simple, it would make sense to consider early prediction-based feedback similar to legacy feedback for encoding purpose at the user. Therefore, the user can apply the same encoding and transmit processing for the early prediction-based feedback as of the classical HARQ feedback.

The activation of the two-stage feedback can be done from the gNB for the sensitive URLLC traffic when the gNB considers that latency budget is critical to meet the reliability target and may activate the two-stage feedback for such URLLC traffic. This activation can be sent in the higher layer signaling to the user.

If there are some services where dynamic control is necessary for feedback purpose or in case of sporadic traffic instants when in some occasions latency budget may be very critical, it will be advantageous to have the dynamic control over the nature of the feedback. For such cases, having an indication in the DCI for the user regarding activation of two-stage feedback will be very helpful. In one method, this indication can be a single bit flag which indicates the activation/de-activation of the two-stage feedback. The resources where UE may potentially transmit early prediction-based feedback upon activation might have been pre-assigned to the UE in the higher layer signalling.

Figure 8 shows a downlink transmission with early feedback and the gNB sensing latency budget in accordance with embodiments of the present invention. As can be seen in figure 8, in the first transmission, the gNB sends PDCCH and PDSCH (C1 and D1, respectively) then the UE predicts a failure of the decoder and transmits an early NAK (P1) but the gNB still waits HARQ feedback (F1) to confirm that failure and retransmits the packet (C2 and D2) because it senses the remaining latency budget and recognizes that it still has enough time left to reach the target reliability with the conventional classical HARQ feedback. In the retransmission, the UE continues to predict a failure of the decoder (P2) . This time, the gNB senses that there is no sufficient latency budget left and that, thus, a useful retransmission is impossible in the time constraint if it waits classical HARQ feedback (F2) . For this reason, in case data is actually not decoded correctly, the packet will be lost. Therefore, the gNB reacts very fast to this early feedback to trigger an immediate retransmission (C3 and D3) to increase the chance that the UE can decode data correctly and the reliability of the system is boosted.

In a fifth proposal, the gNB senses latency budget following URLLC requirement so as to decide to use early feedback or HARQ feedback. The gNB only uses early feedback when latency budget does not remain enough to wait for the HARQ feedback in order to make a decision about whether or not to trigger a retransmission. This combination brings 2 benefits for the downlink transmission: latency is reduced due to the HARQ RTT in regular HARQ process and increased reliability as described below, and harmful effects of false prediction of early feedback are alleviated when a strategy to choose to use early feedback or regular HARQ feedback is implemented as also described below.

This scheme not only raises reliability of the system but also decreases the influences of false negatives and especially, false positive most of the time. When there is a sufficient latency budget still available, early feedback is not considered by the gNB to decide a retransmission so false prediction has no effect to the system. When latency budget is not enough for regular HARQ feedback, early feedback is used. However, in this case, false positive does not cause the detrimental effect as if early feedback is used at the beginning because the error still happens no matter whether early feedback is used or not, because of the shortage of time to do the retransmission. Resource consumption due to false negative might happen only one time at the end of the transmission and is acceptable under the specific limit when latency and reliability are prioritized in URLLC.

Figure 9 shows how the gNB stops a retransmission triggered by early NAK after receiving ACK HARQ feedback in accordance with embodiments of the present invention. An alternative scheme is also considered when it causes the waste of resources but creates more retransmission occasion than the above scheme. If the gNB receives early ACK, it does not take that feedback into account and continues to wait for HARQ feedback in order to decide to terminate or retransmit data. Therefore, the system avoids suffering from losing packet due to false positive. On the other hand, if the gNB receives early NAK (P1) as in figure 9, it carries out an immediate retransmission (C2 and D2) . After that, if HARQ feedback is NAK, that retransmission still continues. This means that the early retransmission can be translate to more transmission occasions if data continues not to be decoded correctly. In contrast, if HARQ feedback is ACK (F1) , that retransmission is no longer necessary. As illustrated in figure 9, the gNB will stop that retransmission (C2 and D2) instantly after receiving ACK HARQ feedback to prevent from wasting resources and leaves resources for other UEs. These available resources are important in the case of UEs multiplexing.

In a sixth proposal, the gNB only takes early NAK into account. The retransmission is carried out immediately after receiving an early NAK. This retransmission continues if the gNB receives NAK HARQ feedback later. On the other hand, if the gNB receives ACK HARQ feedback that means a false negative of early feedback, the gNB stops immediately the retransmission to reduce to waste resource for the unnecessary retransmission and the transmission completes.

Figure 10 shows a flowchart of the downlink transmission process with early prediction feedback in accordance with embodiments of the present invention. The flow chart of figure 10 illustrates the downlink transmission with two-stage feedback including early feedback based on LLR estimation of a portion of PDSCH as discussed in the fifth proposal. When the transmission starts, the gNB transmits PDCCH and PDSCH in the downlink. The UE receives the signal and starts to estimate and decode them in parallel.

If the UE cannot decode PDCCH, it does not know resource allocation of PDSCH so it is not able to carry out the estimation and decoding. For this reason, no prediction is sent to the gNB. Consequently, the gNB discovers the discontinuous signal and retransmits immediately both PDCCH and PDSCH rather than waiting a regular HARQ feedback.

If the UE decodes correctly PDCCH, it can begin to estimate and decode data. If an ACK prediction is sent to the gNB, the gNB still waits HARQ feedback to confirm the status of the transmission. If HARQ feedback is ACK, the transmission completes. If HARQ feedback is NAK, the gNB will retransmit PDCCH and PDSCH.

If a NAK prediction is sent to the gNB, the gNB will sense latency budget to decide whether it has enough time left to wait HARQ feedback. If the remaining time is sufficient, it waits HARQ feedback and takes an appropriate action based on the value of feedback. If feedback is ACK, transmission is accomplished. If not, the gNB retransmits control and data. On the contrary, if the remains of latency budget are not enough for a useful retransmission following HARQ process, the gNB takes an immediate step to do retransmission to increase the chance of a successful transmission when the UE has one more opportunity to decode data.

In detail, figure 10 illustrates the method 100 of data transmission between a base station gNB and a user equipment UE for both the base station and the user equipment. For the base station, the method comprises the following steps. Typically, the base station is in an idle state 110 when receiving data to be transmitted to the user equipment. In step 120, the data is transmitted to the user equipment using the PDCCH and the PDSCH. In step 140, the base station receives a signal indicative of early prediction feedback from the user equipment and proceeds as described in detail above. Based on the signal indicative of early prediction feedback, if the signal indicative of early prediction indicates DTX (see 152) , the data is immediately retransmitted (see step 120) to the user equipment and the method is repeated again from step 140. If the signal indicative of early prediction indicates NAK (see 154) , the base station determines whether a sufficient latency budget is available as described in detail above. If the budget is available, the base station continues the method 100 at step 166 (see below) of receiving an HARQ feedback. If the budget is not available, the base station retransmits (see step 120 above) the data again to the user equipment and repeats the method from step 140 of receiving a signal indicative of early prediction feedback. In step 166, an HARQ feedback is received (including waiting for the HARQ feedback) . If the HARQ feedback indicates NAK (see 178) , the base station retransmits (see step 120) the data again to the user equipment and repeats the method from step 140 of receiving a signal indicative of early prediction feedback.

For the user equipment, the method 100 comprises the following steps. Substantially corresponding to step 120 above (the base station transmitting the data) the user equipment receives the data from the base station. In step 130, the user equipment estimates an error probability based on at least a portion of the received data. The user equipment then generates a signal indicative of early feedback based on the estimated error probability as described in detail above. Corresponding to step 140 (the base station receiving the signal of early prediction feedback) , the user equipment sends the signal indicative of early prediction feedback to the base station.

Figure 11 shows a prediction error rate in accordance with embodiments of the present invention. In the shown simulation, an input codeword with size 1280 is encoded by base graph 2 of LDPC code as agreed in 3GPP standard. The encoded codeword is modulated by QPSK and transmitted in AWGN channel. The decoder uses message passing algorithm with min-sum calculation to decode the incoming codeword. The maximum iteration of the decoder is 25. The predictor also uses the same algorithm as the decoder but with fewer iterations that is 5. The number of iterations for the prediction estimation can be adjusted to choose the optimal operating point for the trade-off of the prediction-processing-time and prediction reliability. The simulation tests the codeword with rate 1/4 and 1/5. In the predictor, two cases are considered. In the first case, a half of the transmitted codeword is used to estimate the outcome. In the other case, a third of the transmitted codeword is used for prediction. The prediction error including both false negative and false positive is calculated when block error rate (BLER) of the whole codeword is approximate to 10 ^-2. The predictor works much better with a lower rate. The reason is that at a lower rate, the codeword is longer so the portion of the codeword that is used to estimate the error probability is also longer. The predictor has more information and the sub-codeword in the predictor has a higher probability to converge. With code rate 1/5, prediction error rate is less than 0.06, while with code rate 1/4, the prediction error rate is about 0.1. In order to guarantee the error prediction rate below 0.1 when BLER is 10-2 and low processing time of the predictor, the ratio between the code used for prediction and the whole transmitted code should be between 1/3 and 1/2.

Figure 12 shows error rate’s false positives in accordance with embodiments of the present invention, for the same simulation as shown in figure 11. As shown in the graph, the error rate of false positive is much smaller (around 10 times smaller) than the overall prediction error. This means that false negative occurs more than false positive. However, as analysed in part A, false negative is less severe than false positive. Moreover, two proposed schemes in part B also reduce the effect of false negative when the gNB decides to use NAK prediction by sensing latency budget or stops the retransmission after receiving ACK. False positive has harmful effects but the occurring probability is very small. Besides, it also has no influence to the performance of the system in the two proposed strategies.

A small false positive has very positive impact in improving the QoS requirements for the URLLC. As an example, the above figure shows that for a transmission with target BLER of 10 ^-2, the false positive prediction error of 10 ^-3 is achievable. This implies that in general the transmission design is such that the gNB needs to retransmit 1%of the packets. With early prediction, the false positive is 10 ^-3, which means that in the error cases, the early prediction would already have requested the retransmission from the gNB and only in 0.1%of the cases, the gNB would have to trigger the retransmissions after the 2 ^nd stage classical feedback. This certainly comes at the price of the increased resource utilization of the two-stage feedback but for critical URLLC traffic when there are requirements to meet a certain reliability within a certain latency, this can provide very useful means to achieve such targets.

The process in accordance with embodiments of the present invention entail at least the following aspects:

The UE uses a part of the received signal to estimate the error probability of the decoding process based on LLR estimation and generates early feedback to the gNB. The gNB can use early feedback to trigger an immediate retransmission.

Early prediction feedback is used as an indicator of a success or a failure in decoding PDCCH. In case an early feedback is missing, the gNB detects a DTX and starts a retransmission immediately instead of waiting for a normal time interval of HARQ feedback.

The UE can use other metrics from earlier signal as decoded DCI to support the reliability of the prediction.

Threshold in feedback prediction is adapted to control the probabilities of false negative and false positive.

The gNB senses latency budget following URLLC requirement so as to decide to use early feedback or HARQ feedback. The gNB only uses early feedback when latency budget does not remain enough to wait HARQ feedback in order to make a decision about a retransmission.

The gNB only takes early NAK into account. The retransmission is carried out immediately after receiving an early NAK. This retransmission continues if the gNB receives NAK HARQ feedback later. On the other hand, if the gNB receives ACK HARQ feedback that means a false negative of early feedback, the gNB stops immediately the retransmission to reduce to waste resource for the unnecessary retransmission and the transmission completes.

Although not shown in detail any of the devices or apparatus that form part of the network may include at least a processor, a storage unit and a communications interface, wherein the processor unit, storage unit, and communications interface are configured to perform the method of any aspect of the present invention. Further options and choices are described below.

The signal processing functionality of the embodiments of the invention especially the gNB and the UE may be achieved using computing systems or architectures known to those who are skilled in the relevant art. Computing systems such as, a desktop, laptop or notebook computer, hand-held computing device (PDA, cell phone, palmtop, etc. ) , mainframe, server, client, or any other type of special or general purpose computing device as may be desirable or appropriate for a given application or environment can be used. The computing system can include one or more processors which can be implemented using a general or special-purpose processing engine such as, for example, a microprocessor, microcontroller or other control module.

The computing system can also include a main memory, such as random access memory (RAM) or other dynamic memory, for storing information and instructions to be executed by a processor. Such a main memory also may be used for storing temporary variables or other intermediate information during execution of instructions to be executed by the processor. The computing system may likewise include a read only memory (ROM) or other static storage device for storing static information and instructions for a processor.

The computing system may also include an information storage system which may include, for example, a media drive and a removable storage interface. The media drive may include a drive or other mechanism to support fixed or removable storage media, such as a hard disk drive, a floppy disk drive, a magnetic tape drive, an optical disk drive, a compact disc (CD) or digital video drive (DVD) read or write drive (R or RW) , or other removable or fixed media drive. Storage media may include, for example, a hard disk, floppy disk, magnetic tape, optical disk, CD or DVD, or other fixed or removable medium that is read by and written to by media drive. The storage media may include a computer-readable storage medium having particular computer software or data stored therein.

In alternative embodiments, an information storage system may include other similar components for allowing computer programs or other instructions or data to be loaded into the computing system. Such components may include, for example, a removable storage unit and an interface , such as a program cartridge and cartridge interface, a removable memory (for example, a flash memory or other removable memory module) and memory slot, and other removable storage units and interfaces that allow software and data to be transferred from the removable storage unit to computing system.

The computing system can also include a communications interface. Such a communications interface can be used to allow software and data to be transferred between a computing system and external devices. Examples of communications interfaces can include a modem, a network interface (such as an Ethernet or other NIC card) , a communications port (such as for example, a universal serial bus (USB) port) , a PCMCIA slot and card, etc. Software and data transferred via a communications interface are in the form of signals which can be electronic, electromagnetic, and optical or other signals capable of being received by a communications interface medium.

In this document, the terms ‘computer program product’ , ‘computer-readable medium’a nd the like may be used generally to refer to tangible media such as, for example, a memory, storage device, or storage unit. These and other forms of computer-readable media may store one or more instructions for use by the processor comprising the computer system to cause the processor to perform specified operations. Such instructions, generally 45 referred to as ‘computer program code’ (which may be grouped in the form of computer programs or other groupings) , when executed, enable the computing system to perform functions of embodiments of the present invention. Note that the code may directly cause a processor to perform specified operations, be compiled to do so, and/or be combined with other software, hardware, and/or firmware elements (e.g., libraries for performing standard functions) to do so.

The non-transitory computer readable medium may comprise at least one from a group consisting of: a hard disk, a CD-ROM, an optical storage device, a magnetic storage device, a Read Only Memory, a Programmable Read Only Memory, an Erasable Programmable Read Only Memory, EPROM, an Electrically Erasable Programmable Read Only Memory and a Flash memory. In an embodiment where the elements are implemented using software, the software may be stored in a computer-readable medium and loaded into computing system using, for example, removable storage drive. A control module (in this example, software instructions or executable computer program code) , when executed by the processor in the computer system, causes a processor to perform the functions of the invention as described herein.

Furthermore, the inventive concept can be applied to any circuit for performing signal processing functionality within a network element. It is further envisaged that, for example, a semiconductor manufacturer may employ the inventive concept in a design of a stand-alone device, such as a microcontroller of a digital signal processor (DSP) , or application-specific integrated circuit (ASIC) and/or any other sub-system element.

It will be appreciated that, for clarity purposes, the above description has described embodiments of the invention with reference to a single processing logic. However, the inventive concept may equally be implemented by way of a plurality of different functional units and processors to provide the signal processing functionality. Thus, references to specific functional units are only to be seen as references to suitable means for providing the described functionality, rather than indicative of a strict logical or physical structure or organisation.

Aspects of the invention may be implemented in any suitable form including hardware, software, firmware or any combination of these. The invention may optionally be implemented, at least partly, as computer software running on one or more data processors and/or digital signal processors or configurable module components such as FPGA devices.

Thus, the elements and components of an embodiment of the invention may be physically, functionally and logically implemented in any suitable way. Indeed, the functionality may be implemented in a single unit, in a plurality of units or as part of other functional units. Although the present invention has been described in connection with some embodiments, it is not intended to be limited to the specific form set forth herein. Rather, the scope of the present invention is limited only by the accompanying claims. Additionally, although a feature may appear to be described in connection with particular embodiments, one skilled in the art would recognise that various features of the described embodiments may be combined in accordance with the invention. In the claims, the term ‘comprising’ does not exclude the presence of other elements or steps.

Furthermore, although individually listed, a plurality of means, elements or method steps may be implemented by, for example, a single unit or processor. Additionally, although individual features may be included in different claims, these may possibly be advantageously combined, and the inclusion in different claims does not imply that a combination of features is not feasible and/or advantageous. Also, the inclusion of a feature in one category of claims does not imply a limitation to this category, but rather indicates that the feature is equally applicable to other claim categories, as appropriate.

Furthermore, the order of features in the claims does not imply any specific order in which the features must be performed and in particular the order of individual steps in a method claim does not imply that the steps must be performed in this order. Rather, the steps may be performed in any suitable order. In addition, singular references do not exclude a plurality. Thus, references to ‘a’ , ‘an’ , ‘first’ , ‘second’ , etc. do not preclude a plurality.

Although the present invention has been described in connection with some embodiments, it is not intended to be limited to the specific form set forth herein. Rather, the scope of the present invention is limited only by the accompanying claims. Additionally, although a feature may appear to be described in connection with particular embodiments, one skilled in the art would recognise that various features of the described embodiments may be combined in accordance with the invention. In the claims, the term ‘comprising’ or “including” does not exclude the presence of other elements.

Claims

A method (100) of data transmission between a base station (gNB) and a user equipment (UE) , the method comprising:

receiving, at the base station, data to be transmitted to the user equipment;

transmitting (120) the data to the user equipment;

receiving (140) a signal indicative of early prediction feedback from the user equipment;

based on the signal indicative of early prediction feedback:

- if the signal indicative of early prediction indicates a discontinuous transmission, DTX (152) , retransmitting (120) the data to the user equipment repeating the method (100) from the step of receiving (140) a signal indicative of early prediction feedback;

- if the signal indicative of early prediction indicates a negative acknowledgement, NAK (154) , determining, whether a sufficient latency budget is available and:

- if the budget is available, continuing the method at the step of receiving (166) a hybrid automatic repeat request, HARQ, feedback;

- if the budget is not available, retransmitting (120) the data to the user equipment and repeating the method (100) from the step of receiving (140) a signal indicative of early prediction feedback;

receiving (166) an HARQ feedback; and

if the HARQ feedback indicates NAK (178) , retransmitting (120) the data to the user equipment and repeating the method (100) from the step of receiving (140) a signal indicative of early prediction feedback.
The method (100) of the preceding claim, wherein the step of transmitting (120) the data or of retransmitting (120) the data to the user equipment comprises transmitting (120) the data using a physical downlink control channel, PDCCH, and a physical downlink shared channel, PDSCH.
The method (100) of any one of the preceding claims, wherein the step of receiving (166) an HARQ feedback further comprises, if the HARQ feedback indicates ACK (176) and a step of retransmitting (120) the data has been triggered, stopping retransmitting (120) the data.
The method (100) of any one of the preceding claims, wherein the step of receiving (166) an HARQ feedback comprises waiting for the HARQ feedback, if a sufficient latency budget is available.
The method (100) of any one of the preceding claims, wherein the step of determining, whether a sufficient latency budget is available is based on ultra-reliable and low-latency communications, URLLC, requirements; optionally wherein the URLLC requirements include a reliability measure, a latency measure, a channel quality measure, a measure of available resources; and/or a traffic measure.
A method (100) of data transmission between a base station (gNB) and a user equipment (UE) , the method comprising:

receiving, at the user equipment, data from the base station;

estimating (130) an error probability based on at least a portion of the received data;

generating a signal indicative of early feedback based on the estimated error probability; and

sending the signal indicative of early prediction feedback to the base station.
The method (100) of the preceding claim, wherein the step of estimating (130) an error probability is triggered as soon as at least a portion of the data on PDSCH has been received;
The method (100) of any one of claims 6 or 7, further comprising decoding (130) the data received from the base station; optionally wherein the estimated error probability is indicative of an error probability occurring in the step of decoding the data received from the base station.
The method (100) of any one of claims 6 to 8, wherein estimating (130) the error probability comprises:

determining a log likelihood ratio, LLR, for each bit in a codeword of the data as:

where

L _k is the LLR of the k ^th bit in the codeword, b _k is the decoded bit of the k ^th bit in the codeword, and r _k is the received signal of the k ^th bit in the codeword.
The method (100) of the two preceding claims 8 and 9, wherein the steps of estimating (130) the error probability and of decoding (130) the data are performed substantially in parallel.
The method (100) of any one of claims 6 to 10, wherein the error probability of a decoded bit is determined as:

where

L’ _k is the LLR of the k ^th bit.
The method (100) of the preceding claim, wherein estimating (130) the error probability is further based on a block error rate, BLER, as:

where

M is a length of a codeword of the data used to predict the error probability of the step of decoding the data received from the base station.
The method (100) of the preceding claim, wherein estimating (130) the error probability based on the BLER _estimate, further includes setting a threshold (th) and generating the early prediction feedback indicating ACK or NAK as:
The method (100) of the preceding claim and claim 8, further comprising adapting the threshold (th) in order to control a ratio of false positives (FP) and false negatives (FN) ; optionally the false positives being indicative of a situation in which an early prediction indicates correct decoding of a codeword while the step of decoding fails and/or the false negatives being indicative of a situation in which an early prediction indicates incorrect decoding of a codeword while the step of decoding succeeds.
The method (100) of the preceding claim, wherein:

adapting the threshold (th) is based on system level requirements;

the threshold is determined by the user equipment (UE) ; and/or

the threshold is determined based on a command issued by the base station (gNB) .
The method (100) of any one of claims 6 to 15, wherein the step of estimating (130) an error probability is further based on a signal received earlier as decoded downlink control information, DCI.
A base station (gNB) configured to perform the method of any one of claims 1 to 5.
A user equipment (UE) configured to perform the method of any one of claims 6 to 15.