The invention discloses a method for detecting a signal absence by a user equipment (UE), which monitors a channel between the UE and a base station and on which the UE performs blind decoding.

New markets and techniques for e. g. internet-of-things (loT) are currently being developed, new applications like automatic meter reading from the outside of buildings via cellular networks emerge.

On one hand, such applications demand increase in coverage of the cellular networks. On the other hand, network operators are not willing to spend too much frequency spectrum as highly valuable resource and device manufacturers want to reduce costs for such loT devices, which are produced in potentially high volumes.

To meet these requirements, narrow-band ToT standard (NB- ToT) is developed as a derivative of the long-term evolution standard LTE where the maximum bandwidth is very small compared to other cellular systems.

To achieve high coverage while being constraint by heavy limitations of the system bandwidth, repetition and combining is widely deployed throughout NB-IoT standard. Repetition and combining is a widely used technique where signals or information is repeated on the transmitter side. This allows the receiver to combine received signals or information to increase the reliability of the result and perform error correction, e. g. during detection of a signal or decoding of information conveyed in a certain channel. Through these mechanisms, NB-IoT enhances the link budget by an additional 20dB margin compared to LTE (Long Term Evolution), WCDMA (Wideband Code Division Multiple Access) or GPRS (General Packet Radio Service) to achieve a maximum coupling loss (MCL) of 164dB. The standard also emphasizes on the device being power efficient because the application such as smart metering demands battery lifetimes of more than 10 years.

In cellular communication, the scheduling information of the downlink or uplink data on the data channel is signaled in a so-called Narrowband physical downlink control channel (NPDCCH) where all the user equipments (UE) connected to the base stations (eNB or eNodeB) of the network can tune in. It is scheduled on pre-defined time/frequency resources, configured by the base station. UE tunes-in to the corresponding time frequency resources allocated by base station and attempts to monitor and decode the downlink control information in the so-called Narrowband physical downlink control channel (NPDCCH). Particular NPDCCHs are used to signal the presence and parameters of user data transmissions from eNB to a certain UE or a set of UEs or the availability of time/frequency resources and other parameters for transmissions from a certain UE to eNB. This set of parameters is called downlink control information (DCI).

Scheduling of NPDCCH transmissions is organized in so-called search spaces, where each search space comprises multiple NPDCCH candidates, i. e. time/frequency resources individual NPDCCHs can be mapped onto (see Fig. 1). A UE will be configured by eNB to continuously monitor a particular search space. The eNB uses the common search space to transmit one or more DCIs to one or more UEs for paging and random access purposes. Optionally, the random access procedure is part of the connection establishment between UE and eNB. The paging procedure is used to initiate a connection establishment within one or more UEs that are registered within the network but not yet having an active signaling connection. An actual NPDCCH dedicated to that UE will only be present in one of the candidates of the search space when Service Data Units (SDUs) have to be transferred. In all other cases, the monitored candidates and their associated time/frequency resources may contain NPDCCHs dedicated to other UEs attached to the same eNB, interference signals from other eNBs using the same frequency range or no useful signal at all. The method of monitoring, i. e. attempting to decode certain candidates without knowing whether a signal dedicated to the UE is present or not is called "blind search".

A subframe consists of two equally sized slots of 0.5 ms in time direction. Scheduling in downlink is done on subframe basis. In the downlink processing chain of NPDCCH on eNB side, each transport block is cyclic redundancy check (CRC) bits attached and scrambled with the Radio Network Temporary Identifier (RNTI) of the UE, channel coded and rate-matched. The rate-matched block is scrambled, modulated and mapped to a subframe according to aggregation and repetition level of the NPDCCH candidate. This is shown in figure 3.

In order to allow for reception of NPDCCH under extreme conditions (e. g. high MCL due to high attenuation in basements of buildings), NPDCCH can be transmitted in repetitions, which a UE may collect and combine to successfully decode DCI conveyed in NPDCCH. Repetition means that the same DCI is transmitted from eNB in multiple consecutive subframes, allowing UE to collect and combine received signals before attempting to decode DCI. The UE decodes the downlink control information (DCI) by performing blind search within the search space by monitoring multiple candidates represented by aggregation and repetition. A NPDCCH search space contains multiple NPDCCH subframes with an upper limit of Rm _ax . Rm _ax indicates the maximum number of NPDCCH subframes in the NPDCCH search space. Same or similar encoding and repetition schemes apply to other channels in NB-IoT as well, like NB downlink shared channel (NPDSCH).

The UE calculates the cyclic redundancy check value using the received data. Successful decoding is achieved when CRC passes. The cyclic redundancy check passes when the calculated and received cyclic redundancy check values are identical. Decoding attempts can be made after receiving blocks (or subframes) that can be self-decodable. Early termination refers to the event that the rate-matched block could be successfully decoded before all repetition cycles have been received.

Link adaptation is the task of selecting Modulation and Coding Schemes (MCS) on the considered radio link. Within LTE technology, a closed-loop link adaptation is employed in downlink direction between eNB and UE, in which the UE provides feedback to the eNB about the observed link quality that the eNB can use the feedback information to select the MCS for downlink transmissions to the UE. In contrast to LTE, NB-IoT does not define closed-loop link adaptation based on channel state feedback towards the base station. Since the communication session between modem of the UE and network are supposed to be short, a conservative open-loop estimation of the link quality at the base station side has been considered sufficient. On the other hand, this also means that the block error rate (BLER) operating point can be expected to be more conservative, i. e. the base station would provide more redundancy than required in order to account for estimation errors as well as to outweigh the lower frequency diversity due to the reduced system bandwidth of NB-IoT.

The control information for receiving the paging message is transmitted in a search space of type "common search space type-1 (CSS Type 1)". An NPDCCH candidate in CSS Type-1 search space starts at the start of the search space. In addition, the UE monitors for only one grant for paging, meaning CSS Type-1 monitoring can be stopped after the first detected grant. Figure 2 shows the flowchart showing NPDCCH CSS Type-1 monitoring. Every subframe K from K=1 to K=R _inax is searched for the CRC bits. If the CRC passes then the monitoring stops.

The common search space type 1 conveys DCIs for broadcast paging information that is received by many UEs, which are interested in using the paging service provided by the network. The configuration R _max is common to all UEs, and it is chosen by base station according to the link quality between the base station and the UE, which experiences the worst signaling conditions (highest MCL). Thus, for any UE with better signal quality (lower MCL), the decoding success can be achieved before all of _R _max repetitions are received. When the UE is in Idle mode (not connected to the base station), it shall periodically look for paging indications. Paging is a mechanism in which a network tells the UE "I have something for you". For a low throughput, standards like NB-IoT where small infrequent data transmissions and receptions are prevalent, periodic monitoring by receiving all of the Rmax repetitions would cost additional energy and effectively reducing the battery life. While early termination on successful decoding of NPDCCH is key to saving power, many use cases in NB-IoT do not demand frequent mobile terminated data reception. In other words, NPDCCH signal for Paging would frequently be absent and blind decoding do not yield successful decoding because the base station simply do not transmit any control information. A straightforward approach is that all of R _max repetitions needs to be received and blindly decoded to determine the absence of the signal. Receiving all of the R _max repetitions would cost additional energy and effectively reducing the battery life.

It is therefore the objective of the present invention to provide a method, by which a user equipment (UE) can detect whether or not an expected or wanted signal is actually being transmitted from the base station. If it is not transmitted, i. e. the wanted signal is not present, the UE should be able to terminate the downlink reception at an early stage.

The objective will be solved by a method according to independent claim 1. The inventive method for detecting a signal absence by a user equipment (UE), which monitors a channel between the UE and a base station and on which the UE performs blind decoding, the UE performs the following steps: - step 1: receiving a first subframe repetition of a transmitted signal, storing the first subframe repetition in a repetition-combining buffer and initializing a previous metric P _m to zero;

- step 2: receiving a next subframe repetition of the transmitted signal and calculating a current metric C _m according to by a cross- correlation, with r being soft-bits of a received soft-bit sequence R _n and k is a length of said sequence R _n ;

- step 3: comparing the current metric C _m with the previous metric P _m and if P _m > C _m , increasing a frequency of occurrence f by 1 and setting P _m = C _m -i;

- repeating step 2 and step 3 until the occurrence f exceeds a predefined frequency of occurrence threshold T _f and early terminate a reception of the transmitted signal.

The central idea of the proposed signal absence detection method is to detect the absence of data resource elements on a channel on which the UE performs blind decoding. On detecting the absence, it is concluded that the base station is not transmitting the signal of interest and, thus, the UE performs early termination and therefore power consumption is reduced.

In the NB-IoT network, the UE must synchronize to the base station (eNB or eNodeB), acquire basic information and connect to the base station in order to initiate the data transfer. During this process, information is exchanged between UE and eNB via downlink Narrowband physical downlink shared channels (NPDSCH) and uplink Narrowband physical uplink shared channels (NPUSCH). The downlink control information (DCI) pertaining to uplink (DCI Format NO) and downlink (DCI Format Nl) data transmission is signaled to UE by eNB via NPDCCH.

DCI as NPDCCH payload is 23 bits long and carries the configuration parameters of downlink/uplink data channels such as modulation and coding scheme, resource assignment, repetition number etc. As per 3GPP 36.211 and 36.212, the processing of NPDCCH on the transmitter is as follows:

• Calculation and attachment of 16-bit CRC, resulting in a bit sequence

• Rate 1/3 Convolutional encoding with tail-biting, resulting in a bit sequence

• Rate matching (to the size of a single subframe)

• Scrambling

• QPSK mapping

• Resource mapping

This is shown in figure 3.

The resource-mapped subframe is repeated for R repetitions based on the chosen candidate. Repetition level of candidate is chosen by eNB based on the channel conditions to ensure successful decoding in the UEs as well as minimum radio resource utilization. In NB-IoT, HARQ (Hybrid automatic repeat request) with soft combining is used to increase the decoding performance. During a repetition-combining process, the incorrectly decoded blocks are not discarded but stored in the memory, which is called repetition-combining buffer. An incorrectly decoded block is a block where the UE- calculated CRC does not match the received CRC. On the next retransmission, the soft-bits (LLR- log likelihood ratio) of the retransmitted block are combined with the contents of the repetition-combining buffer.

In a typical NPDCCH channel, the transport block is same, which means the information and redundancy bits are not changed. Therefore, the receiver uses the maximum-ratio combining to combine the previous soft-bits to the current soft-bits Such combining increases the reliability of the soft-bits by adding extra energy in the signal effectively increasing the energy-per-soft-bit to noise ratio The combined signal is finally decoded to reduce the bit-error-rate.

By monitoring the progression of the soft-bits in the repetition-combining buffer (updated previous metric P _m ), it can be determined if the wanted signal is present or not.

The cross-correlation of the time/frequency offset compensated, channel equalized, demodulated signal with the already available soft-bits in the buffer shall increase when the signal is repeated. When the wanted signal is absent, the received signal has a pure additive white Gaussian noise (AWGN) noise-like signal and hence cross correlation of the repetitions would not increase steadily. The cross-correlation result would fluctuate. This core idea combined with the frequency of fluctuation is used to determine the absence of the signal.

This results from the fact that for an AWGN channel soft-bit (Log-likelihood ratio) is proportional to the received symbol amplitude r _k . It is assumed that symbol d is transmitted and y is received, in conventional BPSK/QPSK (Binary Phase-Shift Keying/ Quadrature Phase-Shift Keying), the LLR can be expressed as ocy, for AWGN. (Eq.1)

Received soft-bit sequence R _n is given by where k is the sequence length. For BPSK, symbols can be directly mapped to soft-bits r _k = y _k .

A metric is defined as a correlation of soft-bits in the repetition combining buffer, before (previous) vs. after (current) repetition combining, r _n (j) ★ r _n -!(j).

The current metric is calculated by for j=0...k-l (Eq.2) and the previous metric subframe is received.

In a variant of the inventive method, the channel monitored by the UE is a Narrowband physical downlink control channel (NPDCCH) scheduled on pre-defined time and/or frequency resources, configured by a base station.

The proposed strategy can be used in any 3GPP system involving resource allocation with repetitions by a base station with a blind decoding in UE with limited CRC bits. Examples are NB-IoT, Cat-M (LTE for Enhanced Machine-type Communications), 5G NR (5 ^th Generation New Radio), NWUS (NB- loT Wake-Up-Signal).

In another variant of the inventive method, the threshold T _f is predefined according to a desired working point on a receiver operating characteristic (ROC) curve, at which the system (connection between UE and eNB) is designed to operate. The ROC curve describes the tradeoff between false positive rate (fall-out) and true positive rate

(sensitivity). Reduced sensitivity results in higher power consumption at the UE whereas higher fall-out leads to retransmissions from eNodeB side reducing the system throughput and lowering the service quality for the UE. The fall-out is a design parameter that allows the equipment vendor to differentiate according to power and reliability. In order not to compromise the general operation of the UE, the fall-out shall be small, e. g. 0.1% or 1%.

In a further variant of the inventive method, the repetition-combining buffer contains incorrectly decoded blocks of the soft-bits of the soft bit sequence R _n .

These blocks in the repetition-combining buffer are used to evaluate and perform a Log-likelihood ratio test to assess the goodness of fit of previous received subframes whose CRC failed and a next/current received subframe. It is tested how a calculated previous metric P _m and a current metric C _m relate to each other.

In another further variant of the inventive method, if the calculated current metric C _m as the result of the cross- correlation is greater than the previous metric P _m stored in the repetition-combining buffer, C _m > P _m , and it increases steadily with each repetition, then the wanted signal is present and transmitted by the base station and received by the UE.

By monitoring the progression of the soft-bits in the repetition-combining buffer, it can be determined if the wanted signal is present or not. The cross-correlation of the time/frequency offset compensated, channel equalized, demodulated signal with the already available soft-bits in the repetition-combining buffer increases steadily when the signal is present and repeated with every repetition. In a variant of the inventive method, if the wanted signal of the monitored channel is absent, said channel contains a pure additive white Gaussian noise-like signal.

The proposed inventive method for detecting a signal absence by a user equipment, UE, which monitors a channel between the UE and a base station and on which the UE performs blind decoding will be explained in more detail using exemplary embodiments .

The appended drawings show

Fig. 1 Scheduling scheme of NPDCCH and NPDSCH;

Fig. 2 Monitoring of NPDCCH CSS Type-1;

Fig. 3 Processing of NPDCCH in downlink transmitter.

Fig. 4 Early Termination for NPDCCH Reception (state of the art);

Fig. 5 Reception of all NPDCCH repetitions when NPDCCH is absent (state of the art).

Fig. 6 Early Termination of Reception when NPDCCH absence is detected (according to the proposed method);

The power consumption of a modem of a UE can be reduced over state-of-the-art approaches by detecting the absence of the wanted signal thereby performing early termination. Figure 4 shows an example, when an eNodeB schedules an NPDCCH, the NPDCCH is present, having 16 repetitions, a receiving UE might apply early termination. In particular, the UE2 in deep coverage needs to receive all 16 repetitions for a CRC pass, whereas the UE1 in good coverage might terminate the reception already after 4 repetitions that translates to an energy saving of about 75%.

When the eNodeB is not scheduling a NPDCCH as shown in figure 5, all UEs need to receive the maximum number of NPDCCH repetitions (32 in the considered case) using state- of-the-art techniques, which is furthermore independent from the coverage level the UE is in.

Using the proposed absence detection method, all UEs can additionally benefit from early termination of the receive operation as soon as they detect the absence of the NPDCCH. The absence detection is shown in figure 6 for both UE1 and UE2. The latter is within deep coverage so that the absence detection requires 19 repetitions, whereas the former UE in good coverage requires only 3 repetitions for that purpose. A UE in good coverage is understood as having low coupling loss; a UE in deep coverage is understood as having high coupling loss. In the terminology of 3GPP NB-IoT, the MCL level of up to 144dB corresponds to normal coverage, whereas MCL ranges from 144dB to 154dB (154dB to 164dB) correspond to extended (extreme) coverage.

Paging in RRC_IDLE mode is a broadcast operation. RRC_IDLE mode means that the UE has no RRC (Radio Resource Control) connection established to the eNodeB. The maximum number of NPDCCH repetitions for RRC_IDLE paging determines the physical coverage of the cell. Thus, the proposed absence detection is of essential importance to UEs in good coverage for achieving good power consumption performance during paging in RRC_IDLE mode.