Resource block candidate selection technique employing packet scheduling in wireless communication systems

The present invention relates to shared channel transmission using packet scheduling ARQ in a mobile / wireless communication system. The present invention is particularly applicable to downlink transmission in an OFDMA (Orthogonal Frequency Division Multiple Access) system using an automatic repeat request.

In wireless communication systems using packet scheduling, at least some of the wireless interface resources are dynamically allocated to different users (mobile stations, MS). The dynamically allocated resources are usually mapped to at least one SDCH (shared data channel), where the SDCH corresponds to, for example, the following configuration:

In a CDMA (Code Division Multiple Access) system, one or more codes are dynamically shared among multiple MSs. Alternatively, in an OFDMA system, one or more subcarriers (subbands) are dynamically shared among multiple MSs.

The combination of the above two configurations is realized in OFCDMA (Orthogonal Frequency Code Division Multiple Access) or MC-CDMA (Multicarrier Code Division Multiple Access) systems, in which multiple codes and subcarriers (subbands) are present. Dynamically shared between MSs.

Figure 1 shows a packet scheduling scheme on a shared channel in a system with a single SDCH. The PHY frame represents the minimum time interval in which the scheduler (PHY / MAC scheduler) performs DRA (dynamic resource allocation) at that time interval. Further, the smallest unit that can be assigned is usually determined by one PHY frame in the time domain and one code / subcarrier / subband in the code / frequency domain. In the following, this unit is referred to as RB (resource block). It should be noted that DRA is performed in the time domain and the code / frequency domain.

The main advantage of packet scheduling is, firstly, the multi-user diversity gain from TDS (Time Domain Scheduling). Assuming that each user's channel state changes over time due to rapid (and gradual) fading, at any time moment, the scheduler has available resources (code for CDMA, subcarrier / sub for OFDMA). Band) can be assigned to users with good channel status. A further advantage is the adaptation of dynamic user transfer speeds. Assuming that the data transfer rate required by each user (the service they are running) changes dynamically over time, the scheduler can vary the amount of resources allocated to each user.

For third-generation CDMA mobile communication systems, packet scheduling is by HSDPA (High Speed Downlink Packet Access), which supports the 3GPP (UMTS) standard, and HDR (High Data Transfer Speed), which supports the 3GPP2 CDMA2000 standard. Introduced by.

In addition to leveraging multi-user diversity with TDS in the time domain, OFDMA can leverage multi-user diversity with FDS (Frequency Domain Scheduling) in the frequency domain. This is because the OFDM signal is composed of a large number of narrowband subcarriers (usually grouped into subbands) that can be dynamically assigned to different users in the frequency domain. This allows the user to take advantage of the frequency selective channel characteristics of multipath propagation to schedule the user to a frequency (subcarrier / subband) where good channel quality can be obtained (multiuser diversity in the frequency domain). ).

For practical reasons, the bandwidth in an OFDMA system is divided into a large number of subbands consisting of a large number of subcarriers. Subbands usually consist of successive subcarriers. However, it may be desirable to construct a subband from dispersed discontinuous subcarriers. The smallest unit that can be allocated to a user, expressed as RB (Resource Block), has a bandwidth of one subband and a duration of one PHY frame (consisting of many OFDM symbols). The scheduler can also assign one or more RBs to a user using multiple continuous or discontinuous subcarriers and / or PHY frames.

For example, for 3GPP LTE (long-term evolution), which is currently being standardized (see, for example, Non-Patent Document 1), a 10MHz system can consist of 600 subcarriers with a 15kHz subcarrier spacing. These subcarriers can be grouped into 24 subbands (each with 25 subcarriers), each subband occupying a bandwidth of 375 kHz. Assuming the PHY frame has a duration of 0.5 ms, the RB has magnitudes over 375 kHz and 0.5 ms.

As can be seen from the above, in order to utilize multi-user diversity in the frequency domain to obtain scheduling gain, one user's data is allocated on several RBs that improve the user's channel state. Since such RBs are usually close to each other, this transmission mode is referred to below as LM (localization mode). An example of LM is shown in Fig. 2.

In contrast to LM, OFDMA also allows decentralized allocation of resources in the frequency domain. This is referred to below as DM (Distributed Mode). DM can be achieved in several different ways. For example, a user (code block) is assigned to multiple distributed RBs, subcarriers or modulation symbols, and these RBs are shared by multiple DM users. In addition, a user (code block) can be assigned to multiple distributed subcarriers or modulation symbols, such as those placed in a hole in an RB that is also used by LM.

Transmission in DM is due to, for example, CQI (Channel Quality Indicator) feedback being restricted or degraded, and / or due to overdue CQI feedback (eg, due to high Doppler). It may be effective when the channel quality to the mobile station (each receiver) is not sufficiently grasped on the base station (transmitter) side. Yet another situation in which DM can be used is when the data to be transmitted is likely to reach a delay limit and transmission needs to be enhanced by the use of frequency diversity.

In any PHY frame, whether LM or DM, multiple sign blocks (transport blocks in 3GPP terminology) are placed on several different RBs that belong to or do not belong to the same service or ARQ process. It may be noted that it can be assigned to the same user. From a scheduling or DRA perspective, this can be understood as assigning different users.

In the following, we will focus on OFDMA's LM (localization mode), which typically maps code blocks to a single or multiple contiguous RBs. But without loss of generality, the same is true for DM and other transmission modes or access schemes (eg, CDMA).

To take advantage of scheduling benefits, packet scheduling is typically combined with fast LA (link adaptation) techniques such as AMC (Adaptive Modulation and Coding) and ARQ (Automatic Repeat Request). In addition, high speed and / or loose power control can be applied.

Using AMC (see Non-Patent Document 1), each scheduled data transfer rate per user code block (ie, per PHY frame) was assigned by varying the MCS (modulation and coding scheme). Dynamically adapted to the momentary channel quality of the resource. Of course, this requires an estimate of the channel quality on the transmitter side for the link to each receiver.

In the case of OFDMA, the MCS is adaptable per code block (in the time and / or frequency domain), which can span multiple RBs, and per RB.

ARQ is commonly used to improve the robustness of packet data transmission and to recover from transmission errors caused by inadequate AMC operation. ARQ introduces time diversity for transmission.

A common technique for error detection / correction is based on an ARQ scheme that combines FEC (forward error correction) called HARQ (Hybrid ARQ). When a CRC (Cyclic Redundancy Check) detects an error in a packet, the receiver requires the transmitter to send additional information (retransmission) to increase the probability of correctly decoding the error packet. ..

The packet is encoded by FEC before transmission. Three types of ARQ schemes are defined, depending on how the content of the retransmission and the previously transmitted information are bit-synthesized.

Type I: Received packets with errors are dropped, a new copy of the same packet is retransmitted, and decrypted separately. The version received before the packet and the version received after it are not combined.

Type II: Received packets containing errors are not discarded but are stored at the receiver and decrypted after additional retransmissions and synthesis. The retransmitted packet can contain additional redundant bits (to reduce the effective code rate) and may contain (partially) the same bits as the previous transmission to increase the reliability of the transmit bits. It can, or can include a combination of additional redundant bits and retransmission bits. Note that the modulation scheme, code rate and / or packet size can be changed between retransmissions. HARQ Type II is also known as Increased Redundancy HARQ.

Type III: This type is cited as a special case of Type II with the constraint that each retransmission is here self-decryptable. This means that the transmitted packet can be decrypted without being combined with the previous transmission. This is useful when some transmissions are so compromised that the information is almost completely non-reusable. If all transmissions carry the same data, this is seen as a special case called HARQ Type III with a single redundancy version (or chase synthesis).

The ARQ (HARQ) protocol can be implemented synchronously or asynchronously. In asynchronous ARQ mode, retransmissions can be assigned to any RB and have no temporal relationship to the previous transmission, that is, the transmitter retransmits at any time after receiving a NACK (or an ACK timeout occurs). Can be scheduled. Therefore, in the case of OFDMA LM, TDS and FDS can be used for retransmissions to obtain a multi-user diversity scheduled gain in the time and frequency domains. Figure 3 shows an example of asynchronous ARQ. Note that this figure and subsequent figures show the transmission of only one packet to a single user for illustrative purposes.

In the synchronous ARQ mode, retransmission occurs based on a predetermined timing relationship with the previous transmission and a predetermined RB. The predetermined RB may be the same as the RB of the previous transmission, or may be an RB defined by a pattern. That is, the retransmission is not scheduled and no scheduling gain is obtained. Figure 4 shows an example of a synchronous ARQ in which the RTT (round time) of the ARQ is a 4PHY frame and the same RB is used for retransmission.

Layer 1 and Layer 2 (L1 / L2) control signaling with one or more SDCHs (shared data channels) to inform scheduled users of user allocation status, transmission formats, and data-related parameters. Must be sent.

In 3GPP HSDPA (CDMA), L1 / L2 control signaling is transmitted over multiple SCCHs (shared control channels) in each PHY frame (TTI, 2ms). Each transmit SCCH includes channelization code sets, modulation schemes, transport block size information, redundancy and constellation versions, HARQ process information, new data indicators (similar to HARQ sequence numbers) and user identification. Carry information about one scheduled user.

In general, the information sent by L1 / L2 control signaling can be divided into the following two types.

The SCI (Common Control Information) portion of L1 / L2 control signaling contains information related to resource allocation. Therefore, it is assumed that all users can decrypt the SCI. The SCI usually contains information about user identification and RB assignment.

Depending on the settings of other channels and DCI (Individual Control Information), SCI can be used for ACK / NACK for uplink transmission, MIMO (Multiple Input Multiple Output) related information, Uplink Scheduling Information, DCI (Resource, It can further include information such as information about (MCS, etc.).

The DCI portion of the L1 / L2 control signaling contains information related to the transmission format and the data transmitted to a particular scheduled user. That is, the DCI needs to be decrypted only by its scheduled user. The DCI usually contains information about the modulation scheme and transport block size (or code rate).

Depending on the full channel configuration, SCI format and HARQ settings, DCI can provide information related to HARQ (eg, HARQ process information, redundancy and constellation versions, new data indicators), MIMO related information, etc. Further can be included.

L1 / L2 control signaling can be transmitted in various formats.

The first possibility is to encode SCI and DCI together. For multiple users (code blocks), each SCI and DCI is encoded together. For a single user, its SCI and DCI are encoded together and sent individually for each user.

The second possibility is to code SCI and DCI separately. Therefore, each SCI (or DCI) for multiple users is encoded together, or each SCI or DCI is encoded for each user.

If there are multiple SCI code blocks (each SCI code block can contain its own SCI for multiple users), multiple SCI code blocks are transmitted with different powers, modulations, coding schemes and / or code rates. It is possible (see Non-Patent Document 2).

From a logical point of view, the L1 / L2 control signaling composed of SCI and DCI can be seen as follows, for example. The first option is to have a single (shared) control channel divided into two parts (SCI and DCI). Alternatively, there is a single (shared) control channel (SCI), and the DCI is not a separate control channel, but is part of the SDCH, ie, mapped with the data (same RB). In addition, there can be two separate control channels (SCI, DCI), or multiple separate control channels, such as a single SCI control channel and multiple DCI control channels, multiple SCI control channels. And there may be multiple DCI control channels, or there are multiple SCI control channels, and the DCI is not a separate control channel, but is part of the SDCH, i.e. mapped with the data (same RB). ).

For illustrative purposes only, the following description of this is that if SCI and DCI are encoded separately, then DCI is encoded per user (SCI is encoded per user or collectively for multiple users). If possible, DCI will focus if it is mapped with the data (same RB).

Normally, the SCI is mapped separately from the SDCH in the physical resource, while the DCI can be mapped separately from the SDCH or to the resource assigned to the SDCH. In the following, the latter case will be illustrated in FIG. Here, DCI is mapped to the beginning of the first assigned RB.

Furthermore, as an example for explanation, assume an OFDMA system with SDCH (shared data channel) that uses TDS (time domain scheduling) and FDS (frequency domain scheduling) for LM (localized mode) transmission. To do.

As shown in FIG. 4, the retransmission in the synchronous ARQ operation occurs after a predetermined timing from the previous transmission on the predetermined RB. This means that retransmissions are not scheduled and generally do not require L1 / L2 control signaling (SCI and DCI) for retransmissions. In order to adapt the transmission format (eg, modulation scheme, code rate, code block size, redundancy version) for the retransmission, it may be advantageous to still transmit the DCI during the retransmission. It should be noted that. The advantage of eliminating / reducing L1 / L2 control signaling for retransmission comes at the expense of losing the scheduling gain of TDS and FDS in retransmission. Figure 6 shows a case where the retransmission also includes DCI.

As illustrated in Figure 3, retransmissions in asynchronous ARQ operation are clearly scheduled in the time and frequency domains to obtain scheduling gain from multi-user diversity not only during initial transmission but also during retransmission. .. This has the disadvantage that L1 / L2 control signaling must be transmitted for each retransmission, i.e. SCI and DCI must be transmitted. For example, if 20% of all transmissions in the system are retransmissions, the overhead of L1 / L2 control signaling will increase by the same amount as above compared to synchronous ARQ operation. In addition, asynchronous ARQ reduces the potential for DRX (discontinuous reception) because the user (receiver) will receive the retransmission at any time. This reduces the potential for power savings. FIG. 7 shows an example.

<p num="0044"><nplcit num="1"><text>3GPP, Technical Report 25.814; Physical Layer Aspects for Evolved UTRA, v. 1.0.3, February 2006</text></nplcit><nplcit num="2"><text>3GPP, E-UTRA downlink control channel structure and TP, R1-060378, February 2006</text></nplcit></p>

<p num="0045"> In view of the shortcomings of conventional communication systems, an object of the present invention is the transmission and decoding of data packets, which increases the scheduling gain in the time, code or frequency domain and reduces the amount of signaling data for transmission. To provide a method. This object is solved by the method described in the independent claims.</p><p num="0046"> A further objective is to provide a corresponding transmitter and receiver as well as an improved communication system. To achieve this object, the present invention provides transmitters, receivers and communication systems as defined by independent claims.</p>

<p num="0047"> The proposed solution offers the advantage of asynchronous ARQ, which provides scheduling gain on retransmission while reducing the control signaling required for retransmission. This is achieved by specifying multiple resource candidates for retransmission credit, and the resources actually used are detected semi-blindly.</p><p num="0048"> Therefore, the present invention is based on the idea that resource block candidates are used for transmission and reception, and these RB candidates are semi-blindly decoded to detect data packets destined for the self-receiver.</p><p num="0049"> According to one preferred embodiment, the resource block candidate includes individual control information that allows the receiver to distinguish between data packets destined for its own device and packets destined for another user.</p><p num="0050"> According to yet another preferred embodiment, the individual control information includes a signature or personal identification. According to another preferred embodiment, resource block candidates are either preset or determined by a feedback signal received from the user.<u style="single">According to yet another preferred embodiment, the methods of the invention are dynamically allocated resources in at least one or a combination of time, code or frequency domains on a shared channel of a wireless communication system. A method of decoding a data packet received via a block, in which a step of obtaining information on a plurality of resource block candidates for expected reception of a data packet addressed to a certain receiver and a step of obtaining information on the receiver. It has a step of decoding the resource block candidate in order to detect the addressed data packet.</u><u style="single">According to yet another preferred embodiment, the receiver of the present invention is a plurality of dynamically allocated resources in at least one or a combination of time, code or frequency domains on a shared channel of a wireless communication system. -A receiver for receiving data packets via blocks, a means (310) for obtaining information on multiple resource block candidates for expected reception of data packets, and the receiver. It includes a decoder (320) for decoding the resource block candidate in order to detect the addressed data packet.</u></p><p num="0051"> Alternatively, RB candidates can be selected based on the previous transmission.</p><p num="0052"> In short, the invention can be viewed as a "hybrid" or "soft" solution between synchronous and asynchronous ARQ transmissions. For conventional asynchronous ARQ operation, the present invention maintains similar scheduling gains in time and frequency or code domain without requiring transmission of the SCI portion of control signaling upon retransmission. In contrast to synchronous ARQ operation, the present invention acquires scheduling gain in time and frequency or code domain.</p><p num="0053"> The present invention will be further understood from the following description of preferred embodiments with reference to the accompanying figures.</p>

<figref num="1">SDCH-Shows an example of packet scheduling that multiplexes four mobile stations on a shared data channel.</figref><figref num="2">An example of TDS and FDS in the localized mode is shown.</figref><figref num="3">An example of asynchronous ARQ is shown.</figref><figref num="4">An example of synchronous ARQ when RTT is a 4PHY frame is shown.</figref><figref num="5">An example of SCI and DCI control signaling mapping where DCI is mapped to the data part is shown.</figref><figref num="6">An example of SCI / DCI signaling for synchronous ARQ is shown.</figref><figref num="7">An example of SCI / DCI signaling for asynchronous ARQ is shown.</figref><figref num="8">A flowchart illustrating a transmission method according to an embodiment of the present invention is shown.</figref><figref num="9">A flowchart illustrating a decoding method according to an embodiment of the present invention is shown.</figref><figref num="10">An embodiment of a base station and a mobile station according to the present invention is shown.</figref><figref num="11">An example including SCI / DCI signaling in the case of asynchronous ARQ operation using the principle of the present invention is shown.</figref><figref num="12">An example of resource block candidate specification based on mobile station feedback is shown.</figref><figref num="13">An example of local / proximity RB candidates is shown.</figref><figref num="14">An example of distributed RB candidates is shown.</figref>

FIG. 8 shows a flowchart illustrating a transmission method according to an embodiment of the present invention.

In step 110, resource block candidates for the expected transmission of data packets to the receiver are selected. In at least one or a combination of time, code or frequency domains, preferably in time / frequency or time / code domains, multiple resource blocks are dynamically allocated.

After selecting a candidate, a data packet is sent using at least one assigned RB from the RB candidates selected in step 110 (step 120).

Finally, the transmitter receives feedback information from the receiver regarding the transmitted data packet, such as an acknowledgment or negative response message or channel quality information.

Based on the feedback message, the transmitter can adapt or modify the selection method performed in step 110 for resource block candidates for the next transmission. In the ARQ system, the next transmission can be the retransmission step of at least a portion of the previously transmitted data packet.

It must be apparent to those skilled in the art that it is not always necessary to change the candidate selected for each transmission and that the selection may be made "as needed".

In addition, it will be understood from the following description that the selection is alternative based on the previous transmission or preset scheme.

In addition, it is understood that the receiver does not need to send feedback regularly, i.e. for each subsequent transmission.

FIG. 9 shows the operation steps of the decoding method in the receiver according to the preferred embodiment.

In step 150, information on resource block candidates for expected reception of data packets is obtained. Candidates can be preset by the network side for individual users or receivers, or can be determined based on communication links, such as measurements or other link parameters.

In step 160, in order to detect data packets destined for the receiver, the receiver decodes all resource block candidates, preferably in parallel, or preferably sequentially in order of priority.

At step 170, the receiver sends feedback to the transmitter to optimize the next transmission. Feedback can be interpreted by the transmitter as instructions or recommendations for how to select resource block candidates.

FIG. 10 shows a preferred embodiment of a transmitter realized as a base station of a wireless communication system and a receiver realized as a mobile station.

Base station 200 includes a resource allocation unit 210 for dynamically allocating resource blocks. To select resource block candidates, the base station includes selection unit 230. In addition, a conventional modulator and encoder 220 maps data packets to resource blocks allocated for transmission over a wireless link to the receiver. The feedback evaluation unit 240 is also included in the base station in order to receive feedback from the receiver as described above. All of the above functional units are interconnected by a conventional data bus and are under the control of a central control unit, not shown for simplicity.

Corresponding functional components of the mobile station 300 include a demodulator and decoder 320, a resource block candidate information unit 310. The information unit receives information on resource block candidates selected from the base station or the network side. Alternatively, the information unit determines this information itself. In addition, the measurement and control unit 330 can perform the required communication link measurements and detect data packets destined for its own station. Finally, the feedback signal generation unit 340 provides the base station with feedback signals such as an acknowledgment, a negative response or channel quality information. The feedback signal can also include information about resource block candidate selection.

Time of ARQ retransmission-Position in the frequency grid (RBs arranged on different PHY frames) specifies several RB candidates. Since the user (receiver) generally knows whether or not a retransmission will occur (by AK / NACK transmission), the user attempts to decrypt each DCI that is expected to be transmitted on multiple RB candidates and retransmits. Find the RB whose transmission was actually scheduled. This can be seen as a kind of semi-blind detection of retransmissions, i.e., semi-blind detection of DCI associated with retransmissions. In general, the number of RB candidates is significantly smaller than all theoretically possible RBs for asynchronous ARQs. An example is shown in FIG.

As mentioned above, according to this preferred embodiment, DCI is assumed to be transmitted on the same resource (RB) as the data. Therefore, if the user identifies the DCI directed to him on one of the RB retransmission candidates, the user can detect the retransmission data and attempt to decrypt the data. This is because RB candidates that are not scheduled to be retransmitted to the own machine are considered to be used for data to other users, so the receiver sets the DCI (correct DCI) for the own machine to other devices. It means that it can be determined from the DCI for the user. The correct DCI can be identified, for example:

The DCI conveys the signature or identity, based on which the receiver can detect the DCI directed at it. This signature or identity may be transmitted explicitly (transmitted data bits) or implicitly (for example, as in the case of 3GPP HSDPA, the data portion of DCI is of the user. Scrambled / masked / colored by signature or identity).

Alternatively, DCI does not convey a unique signature or identity, but the receiver can detect DCI directed at it based on at least one of modulation, coding scheme or code rate as a communication link characteristic. ..

In addition, detection can be performed based on DCI data content, such as transport block size, HARQ or MIMO parameters.

It is pointed out that retransmissions can be mapped to multiple RBs, that is, each RB candidate basically corresponds to multiple RBs. The DCI may then be mapped to the RB candidate or may be mapped according to a known format starting with the RB candidate.

RB candidates for retransmission can be specified according to the policy described below. Here, the policy focuses on the case where all RB candidates are in the same PHY frame (see also parts 12-14). However, without loss of generality, the above policy is also applicable to the more general case where RB candidates can span different PHY frames.

RB candidates are feed from the user (receiver) is specified based on the back. In one case, feedback can be transmitted regularly, eg, with a defined duty cycle. In another case, the feedback can also be triggered by the transmitter (base station), where the trigger is the actual data transmission (eg, packet), even if it is an explicit feedback request by the transmitter. May be caused by (transmission of). This feedback may be sent explicitly or implicitly along with CQI (Channel Quality Indicator) feedback. The actual designation of the RB candidate may be instructed directly by the user as a mandatory instruction to the base station, or the user feedback may be interpreted as a recommendation that the base station's RB candidate is determined based on it. Please note that. In this case, the base station needs to notify and signal the user of the RB candidate. The following examples can be given.

The user can feed back the CQI for only the selected (best) RB. The RB candidate is then based on this selected RB. An example is shown in FIG.

Alternatively, the user can feed back a bulk CQI for only multiple RBs. The RB candidate is then based on this selected RB.

As a variant, the compressed CQI of multiple RBs can be fed back at any time, for example by transformation in the time domain by DCT (Discrete Cosine Transform). RB candidates can then be specified based on the best RB (after reconfiguring the CQIs of multiple RBs at the base station).

Further, as an alternative to the above description with respect to FIG. 8, RB candidates may be specified depending on the RB (s) scheduled for the previous transmission (retransmission) or initial transmission. In general, packet transmissions can be mapped to a single or multiple RBs. This mapping can be locally concentrated or distributed. Generally, this needs to be preset by the network side and notified to the mobile station. Here are some examples (not comprehensive):

a) When the initial transmission or previous transmission (retransmission) was transmitted on RBk, the retransmission is within RBk ± n, that is, near the initial transmission or previous transmission (retransmission) (local band, i.e. localization). ) Is sent. An example is shown in FIG.

b) When the initial transmission or the previous transmission (retransmission) was transmitted on the RB from RBk to RBk + l, the retransmission is the RB between RBk-m and RBl + m, that is, the initial transmission or the previous transmission. It is transmitted near (local band) near transmission (retransmission).

c) When the initial transmission or the previous transmission (retransmission) was transmitted on RBk, the retransmission is RB of RBk ± m · n (m&gt; 1, n = 1, 2, ..., N), that is, , Is transmitted in RB distributed within a certain band relative to the initial transmission or the previous transmission (retransmission). An example is shown in FIG.

d) When the initial transmission or the previous transmission (retransmission) is transmitted on RBk to RBk + l RBs, the retransmission is RBk ± m · n (m&gt; 1, n = 1, 2, .. ., N) RB, that is, RB distributed within a certain band relative to the initial transmission or the previous transmission (retransmission).

e) Total system bandwidth into M frequency blocks, each frequency block N_mIt can be divided to include RB of 1. When the initial transmission or the previous transmission (retransmission) is transmitted on the frequency block m, the RB candidate of the retransmission credit can be specified only on the frequency block m. Alternatively, the retransmission RB candidate can be specified on another preset frequency block with p m. Furthermore, all RBs in the selected frequency block can be candidates, or are limited to those selected according to each of the above examples.

If the system bandwidth is greater than the bandwidth performance of several mobile stations, splitting into frequency blocks is useful. In this case, several mobile stations can only receive the selected frequency block. For example, when the system bandwidth is 20MHz and it is divided into 4x5MHz frequency blocks, only mobile stations with 5MHz bandwidth performance can receive one of the four frequency blocks and also at 10MHz. A mobile station with bandwidth performance can receive two adjacent blocks out of four frequency blocks. Naturally, a mobile station with a bandwidth performance of 20 MHz can receive all frequency blocks.

Further, as an alternative to the above description with respect to FIG. 8, RB candidates can be pre-specified regardless of the RB to which the initial transmission or the previous transmission (retransmission) was transmitted. Generally, this needs to be preset by the network side and notified to the mobile station. For example a) RB candidates are preset in a distributed manner, preferably within a band that the mobile station can receive.

b) RB candidates are locally preset, preferably within a band that the mobile station can receive.

c) RB candidates are in preset frequency blocks.

In addition to the general RB candidate settings described above, this preferred embodiment is the case where all RB candidates for any retransmission credit are in the same PHY frame. Examples are shown in FIGS. 12 and 14.

Compared to common solutions, this has the potential for efficient DRX (discontinuous reception) operation. Furthermore, this is realized in the same way as synchronous ARQ in that the retransmission has a certain timing relationship with the previous transmission, that is, the timing of the RB candidate is known to the mobile station, and the RB candidate is specified by the frequency domain. can do. Note that in this case the scheduling gain on retransmission is limited to the FDS, but this gain is usually sufficient.

Hereinafter, variations of the above-described exemplary embodiments will be described, which will be apparent to those skilled in the art.

As a final variant, the DCI on retransmission can be adapted by changing the power from the initial transmission. It is advantageous to increase or decrease the power, depending on other configuration parameters. Furthermore, in retransmission, the DCI size may be different from that at the time of initial transmission.

Depending on the specified RB candidate policy, RB candidates for different users may have the following additional characteristics:

a) Minimize duplication of RB candidates. In this case, the RB candidates are specified so that the RB candidates of different users are not the same. This avoids retransmission conflicts and resource shortages in retransmission credit. If the total number of available RBs is less than the total number of RB candidates, some of the RB candidates for different users may be duplicated. This can be avoided, for example, by resetting the RB candidates so that the total number of RB candidates is less than the total number of RBs. Alternatively, it is possible to specify RB candidates so that the amount of duplication is the same for all users, even if there is minimal duplication.

b) Make as much duplication of RB candidates as possible for different users. This can be seen as a virtual retransmission channel where a kind of statistical multiplexing of retransmissions occurs on the channel. For example, if there are N users in the system and M RB candidates are specified for each user, these M RB candidates can be the same for all users. Note that this is not possible in all cases, as some users may be assigned to different frequency blocks.

c) RB candidates for different users partially overlap. For example, there are N users in the system, and M RB candidates are specified for each user. If M N, then these M users can have the same RB candidates, that is, the M users share a kind of virtual transmission channel.

As yet another variant, this concept can work as well in cases where DCI is not mapped to the RB where the data is scheduled. In this case, DCI shall indicate the location of the corresponding data portion implicitly or explicitly.

The method of the present invention may already be carried out not only for retransmission but also for initial transmission. That is, the mobile station has already detected the DCI of the initial transmission semi-blindly, and no SCI is required at all.

Semi-blind detection of DCI on RB candidates is typically done in parallel or sequentially. For sequential semi-blind detection, for example, the mobile station may start with an RB candidate with the best actual or reported channel quality, or the mobile station may be the initial transmit or previous transmit (retransmit) RB. Mobile stations may use an efficient method for ordering RB candidates according to priority, such as starting with the RB candidate closest to.

Similar to the preferred embodiments shown in FIGS. 12-14, the RB candidates can be the same RBs in the frequency domain, in this case specified in the time domain. This allows TDS for retransmissions, but not FDS. Compared to preferred embodiments, this has the disadvantage of being out of time synchronization and incurring additional delays.

In a Type I HARQ scheme, this method works well with semi-blind detection of retransmission data without a DCI of retransmission credit. Theoretically, these measures work for Type II / III HARQ schemes as well, but receive because the receiver needs to store the data of all received RB candidates for all failed retransmissions. The HARQ buffer required by the machine becomes large. Moreover, in the worst case, the receiver will have to attempt to synthesize and decrypt all combinations of RB candidates over multiple retransmissions.