WO2022117187A1

WO2022117187A1 - Multi-cell coordinated precoding

Info

Publication number: WO2022117187A1
Application number: PCT/EP2020/084437
Authority: WO
Inventors: George Koudouridis; Chrysovalantis KOSTA; Anders Dahlen; Per TENGKVIST
Original assignee: Huawei Technologies Co., Ltd.
Priority date: 2020-12-03
Filing date: 2020-12-03
Publication date: 2022-06-09
Also published as: EP4197112A1; CN116458080A

Abstract

The invention relates to a first network node (100) and a second network node (300) for multi-cell coordinated precoding. The first network node (100) determines a first set of client devices (600a, 600b,…, 600n) served by the first network node (100), which have correlated channels and are cell-edge client devices in an inter-cell interference area. The first network node (100)5 transmits a channel representation indicating a representation of the correlated channels of the first set of client devices (600a, 600b,…, 600n) to the second network node (300). Based on the received channel representation the second network node (300) determines a precoder for beamforming towards a second set of client devices (600a´, 600b´,…, 600n´) served by the second network node (300). As the precoder is determined taking the first set of client devices 10 (600a, 600b,…, 600n) into account, the inter-cell interference can be reduced.

Description

MULTI-CELL COORDINATED PRECODING

Technical Field

Embodiments of the invention relates to a first network node and a second network node for multi-cell coordinated precoding. Furthermore, embodiments of the invention also relates to corresponding methods and a computer program.

Background

A new radio (NR) massive multi input multi output (MIMO) multi-cell inference suppression technique is to perform coordinated beamforming (CBF). When using CBF, cells calculate precoder in the physical layer taking their own scheduled user equipments (UEs) into account, as well as (cell-edge) UEs in neighbouring cell(s). This can be done by precoding the cells jointly, called joint precoding, or it can be done by other precoding techniques that does not require joint precoding, called multi-cell aware precoding. In a multi-cell aware precoding, the precoder is calculated uniquely per cell, but the scheduled UEs of neighbouring cell(s) are added as interference component to the precoder.

To perform CBF the neighbouring cells share their scheduling results either to a single precoder function or to the precoder functions of the neighbouring cells. Precoding is centralized if the neighbouring cells base band processing are centrally located (e.g. co-hosted in the same location) or if there is a single precoder and the cells base band units are distributed. If the cell base band processing is distributed and the precoder functions of each cell is used then the precoding is distributed.

Summary

An objective of embodiments of the invention is to provide a solution which mitigates or solves the drawbacks and problems of conventional solutions.

Another objective of embodiments of the invention is to provide a solution which reduces overhead signalling and interference in a communication system compared to conventional solutions.

The above and further objectives are solved by the subject matter of the independent claims.

Further advantageous embodiments of the invention can be found in the dependent claims. According to a first aspect of the invention, the above mentioned and other objectives are achieved with a first network node for a communication system, the first network node being configured to serve a first cell using beamforming, and further configured to determine a first set of client devices served by the first cell, wherein the first set of client devices have correlated channels; determine a channel representation indicating a representation of the correlated channels of the first set of client devices; and transmit the channel representation to a second network node.

An advantage of the first network node according to the first aspect is that the channel representation constitutes a scalable and robust solution of channel information exchange for the coordination of the first network node and the second network node to efficiently serve the first set of client devices. This solution reduces significantly the signalling overhead between the first and the second network nodes for the purpose of interference coordination and multiple input multiple output antenna transmission precoding.

In an implementation form of a first network node according to the first aspect, determine the first set of client devices comprises determine the first set of client devices based on a channel correlation distance.

An advantage with this implementation form is that it allows the first network node to flexibly determine the size of the first set of client devices to improve the robustness of the interference coordination between the first and the second network node. The channel correlation distance can be effectively represented with a geographical distance and/or location which allows for determining multiple input multiple output antenna transmission precoding and the resulted beam forming optimally.

In an implementation form of a first network node according to the first aspect, determine the first set of client devices further comprises determine the first set of client devices based on a comparison of the channel correlation distance and a channel correlation distance threshold value.

An advantage with this implementation form is that it allows the first network node to efficiently determine the size of the first set of client devices to improve interference coordination while reducing signaling overhead and the precoding processing time. The threshold may be a geographical distance threshold or a location threshold which allows for determining multiple input multiple output antenna transmission precoding and the resulted beam forming optimally. In an implementation form of a first network node according to the first aspect, determine the channel representation comprises determine the channel representation based on statistics of the correlated channels of the first set of client devices.

An advantage with this implementation form is that it allows the efficient mapping of the correlated channels of the first set of client devices to a channel presentation that effectively represents all client devices in the first set. The channel representation may correspond to a correlated channel of the first set of client devices with the average correlation distance, minimum correlation distance, maximum correlation distance, correlation distance within a high and low correlation distance variance, or may correspond to multiple correlated channels from the correlation distance distribution of the first set of client devices, or any combination thereof. As a result, the channel representation of the correlated channels of the first set of client devices can be effectively used to calculate the weights of the multiple input multiple output antenna precoder in the second network node. These weights determine the transmitting/receiving beams in downlink/uplink so that any interference from the second network node to the first set of client devices is ultimately nullified or significantly reduced.

In an implementation form of a first network node according to the first aspect, determine the channel representation comprises determine the channel representation based on at least one of: client device mobility, client device traffic activity, client device channel stability and client device interference.

An advantage with this implementation form is that it allows to determine the channel presentation based on different optimisation criteria related to client devices of the first set in order to determine the most representative. The channel representation can be optimally selected given the client device and the properties of the correlated channel to improve interference coordination and the effect of multiple input multiple output antenna precoder in the first and second network nodes on the spectral efficiency of the first set of client devices. To this end, the channel representation may effectively correspond to a correlated channel of a client device of the first set of client devices that is stationary or mobile having average, lowest and highest speed and velocity values. In addition, the channel representation may correspond to a correlated channel of a client device of the first set of client devices with certain channel stability as given by average/minimum/maximum coherence time and/or bandwidth, and channel interference, as given by worst interfering channel, or average interfering channel, or geographically being in the centre of the geographical area within the geographical distance. Further, the channel representation may correspond to a correlated channel of a client device of the first set of client devices that has average traffic load, peak traffic load, minimum traffic demands, or traffic of certain quality demands.

In an implementation form of a first network node according to the first aspect, the first network node is further configured to determine a time parameter indicating a valid time period for the channel representation based on at least one of: client device mobility, client device traffic activity, client device channel stability and client device interference; and transmit the time parameter to the second network node.

An advantage with this implementation form is that it allows the first network node to timely follow changes in the first set of client devices in terms of their mobility, traffic activity, channel stability and interference and assist the second network node to timely adapt interference coordination and precoding to these changes and improve spectral efficiency of the first set of client devices.

In an implementation form of a first network node according to the first aspect, the channel representation indicates an identity of a selected first client device in the first set of client devices.

An advantage with this implementation form is that it enables solutions relying on a first set of client devices where the client devices transmits uplink reference signals that can be measured by both the first and the second network nodes.

In an implementation form of a first network node according to the first aspect, the channel representation further indicates a sequence code for the selected first client device in the first set of client devices.

The channel representation may also indicate a frequency pattern and a time pattern for the selected first client device.

An advantage with this implementation form is that it enables the effective and correct decoding and measuring of the uplink reference signal for the correlated channel of the selected first client device using the indicated sequence code at a frequency and a time instance given by the frequency pattern and time pattern, respectively. Effective and correct measurements improve interference coordination and precoding effects on the spectral efficiency of the first set of client devices.

In an implementation form of a first network node according to the first aspect, the channel representation indicates channel coefficient information.

An advantage with this implementation form is that it enables solutions relying on a first set of client devices where the client devices transmits measurements of the first and second network node downlink reference signals that can be measured by the client device and reported back to the first network node.

According to a second aspect of the invention, the above mentioned and other objectives are achieved with a second network node for a communication system, the second network node being configured to serve a second cell using beamforming, and further being configured to receive a channel representation from a first network node, wherein the channel representation indicates a representation of correlated channels of a first set of client devices served by a first cell of the first network node; and determine a precoder for beamforming towards a second set of client devices served by the second cell based on the channel representation.

An advantage of the second network node according to the second aspect is that it allows the efficient mapping of the correlated channels of the first set of client devices to a channel presentation that effectively represents all client devices in the first set of client devices. The channel representation may correspond to a correlated channel of the first set of client devices with the average correlation distance, minimum correlation distance, maximum correlation distance, correlation distance within a high and low correlation distance variance, or may correspond to multiple correlated channels from the correlation distance distribution of the first set of client devices, or any combination thereof. As a result, the channel representation of the correlated channels of the first set of client devices in combination with the correlated channels of the second set of client devices, which are known to the second network node, can be effectively used to calculate the weights of the multiple input multiple output antenna precoder in the second network node. These weights determine the transmitting/receiving beams in downlink/uplink so that any interference from the second network node to the first set of client devices is ultimately nullified or significantly reduced, while the second set of client devices are served highest achievable spectral efficiency. In an implementation form of a second network node according to the second aspect, the second network node is further configured to receive a time parameter from the first network node, wherein the time parameter indicates a valid time period for the channel representation; and perform beamforming in the second cell according to the precoder during the valid time period

An advantage with this implementation form is that it allows the second network node to timely follow changes in the first set of client devices in terms of their mobility, traffic activity, channel stability and interference and assists the second network node to timely adapt interference coordination and precoding to these changes and improve spectral efficiency of the first and second set of client devices.

In an implementation form of a second network node according to the second aspect, the channel representation indicates an identity of a selected first client device in the first set of client devices, and the second network node is further configured to receive reference signals from the selected first client device based on the identity of the selected first client device.

In an implementation form of a second network node according to the second aspect, the channel representation further indicates a sequence code for the selected first client device in the first set of client devices, and the second network node is further configured to decode the reference signals based on the sequence code for the selected first client device to obtain channel coefficient information; and determine the precoder for beamforming based on the channel coefficient information.

An advantage with this implementation form is that it enables the effective and correct decoding and measuring of the uplink reference signal for the correlated channel of the selected first client device using the indicated sequence code at a frequency and a time instance given by the frequency pattern and time pattern, respectively. Effective and correct measurements improve interference coordination and precoding effects on the spectral efficiency of the first and second set of client devices.

In an implementation form of a second network node according to the second aspect, the channel representation indicates channel coefficient information, and the second network node is further configured to determine the precoder for beamforming based on the channel coefficient information.

An advantage with this implementation form is that it enables solutions relying on a first set of client devices where the client devices transmits measurements of the first and second network nodes correlated channels based on downlink reference signals that can be measured by the client device and reported back to the first network node. The channel representation which indicates channel coefficient information is determined by the first network node and transmitted to the second network node which effectively determines the precoder for beamforming that improves interference coordination and consequently the spectral efficiency of the first and second set of client devices.

According to a third aspect of the invention, the above mentioned and other objectives are achieved with a method for a first network node configured to serve a first cell using beamforming, the method comprises determining a first set of client devices served by the first cell, wherein the first set of client devices have correlated channels; determining a channel representation indicating a representation of the correlated channels of the first set of client devices; and transmitting the channel representation to a second network node.

The method according to the third aspect can be extended into implementation forms corresponding to the implementation forms of the first network node according to the first aspect. Hence, an implementation form of the method comprises the features of the corresponding implementation form of the first network node.

The advantages of the methods according to the third aspect are the same as those for the corresponding implementation forms of the first network node according to the first aspect.

According to a fourth aspect of the invention, the above mentioned and other objectives are achieved with a method for a second network node configured to serve a second cell using beamforming, the method comprises receiving a channel representation from a first network node, wherein the channel representation indicates a representation of correlated channels of a first set of client devices served by a first cell of the first network node; and determining a precoder for beamforming towards a second set of client devices served by the second cell based on the channel representation.

The method according to the fourth aspect can be extended into implementation forms corresponding to the implementation forms of the second network node according to the second aspect. Hence, an implementation form of the method comprises the features of the corresponding implementation form of the second network node.

The advantages of the methods according to the fourth aspect are the same as those for the corresponding implementation forms of the second network node according to the second aspect.

The invention also relates to a computer program, characterized in program code, which when run by at least one processor causes said at least one processor to execute any method according to embodiments of the invention. Further, the invention also relates to a computer program product comprising a computer readable medium and said mentioned computer program, wherein said computer program is included in the computer readable medium, and comprises of one or more from the group: ROM (Read-Only Memory), PROM (Programmable ROM), EPROM (Erasable PROM), Flash memory, EEPROM (Electrically EPROM) and hard disk drive.

Further applications and advantages of the embodiments of the invention will be apparent from the following detailed description.

Brief Description of the Drawings

The appended drawings are intended to clarify and explain different embodiments of the invention, in which:

- Fig. 1 shows a first network node according to an embodiment of the invention;

- Fig. 2 shows a method for a first network node according to an embodiment of the invention;

- Fig. 3 shows a second network node according to an embodiment of the invention;

- Fig. 4 shows a method for a second network node according to an embodiment of the invention;

- Fig. 5 shows a communication system according to an embodiment of the invention; - Fig. 6 shows coordinated beamforming according to an embodiment of the invention;

- Figs. 7a-b show coordinated beamforming based on different types of virtual client devices according to an embodiment of the invention; and

- Figs. 8a-b show performance evaluation results of a simulation.

Detailed Description

NR massive MIMO CBF has been designed and verified for centralized solutions. This means that all the important precoding scheduling information of the interfering users/UEs of neighboring cells need to be managed. Therefore, there are information processing, signaling and exchange limitation for the centralized solution as the number of cells and users increase. Although evaluation results demonstrate significant gains in terms of cell throughput and celledge user throughput, such a centralized solution does not allow cells of different sites to be efficiently coordinated if the communication between the cell schedulers and precoders has some delay. Basically, a centralized solution needs to delay transmitting data after scheduling until scheduled result has been communicated to precoder units. Furthermore, the amount of signaling can be large. Centralized CBF is basically designed for coordinating inter-cell interference between cells that has base band processing in the same base band unit, which limits the cells that can be coordinated efficiently.

Solutions have been proposed which attempts to extend the centralize solution by exchanging detailed user scheduling information while trying to coordinate among the users of interfering cells scheduling decisions. Although this is easily done in a centralized deployment of coordinated cells there is no straightforward scalable solution in terms of precoding processing and information exchange on how to perform this in a distributed/decentralized manner.

The solutions for implementing the distributed approach do not effectively cope with the delay in the information exchange between the multiple cells imposed by IP-based communication. Furthermore, to work it requires a considerable amount of information to be exchanged. In principle it requires that extensive and detailed user scheduling information is exchanged among interfering cells. Such an exchange requires both high bandwidth and low latency for the exchange of the information between the cells communicating by means of IP. As a result, the suggested distributed solutions lack efficient inter-cell interference coordination at layer 1 (L1).

An inter-site multi-cell CBF relies on distributed scheduling (DS). In DS, all the cells schedule their users independently, and coordinate scheduled result with each other by exchanging information. The advantage of distributed coordination is that there is no need to deploy a central node, rather expanding on the distributed architecture, which is mostly deployed. However, the main challenge is the information exchange required in terms of delay and bandwidth. Replacing the central solution by a distributed solution would require a distributed precoding function with knowledge about the accumulated scheduled beam vector and the accumulated interfering beam vector at computational time scales much shorter than a time transmission interval (TTI). Assuming that the information is exchanged via the Xn interface, which many times introduces a delay in the order of 5-10ms turns the equivalent distributed solution infeasible. Even in the case where all cells share a common memory, a distributed solution is practically infeasible.

An objective of the invention is therefore to provide an improved CBF which coordinates beamforming at layer 2 (L2) jointly with L1 precoding to minimize inter-cell interference in multicell scenarios and to reduce signaling overhead in the system.

Fig. 1 shows a first network node 100 according to an embodiment of the invention. In the embodiment shown in Fig. 1 , the first network node 100 comprises a processor 102, a transceiver 104 and a memory 106. The processor 102 is coupled to the transceiver 104 and the memory 106 by communication means 108 known in the art. The first network node 100 may be configured for both wireless and wired communications in wireless and wired communication systems, respectively. The wireless communication capability is provided with an antenna or antenna array 110 coupled to the transceiver 104, while the wired communication capability is provided with a wired communication interface 112 coupled to the transceiver 104. That the first network node 100 may be configured to perform certain actions can in this disclosure be understood to mean that the first network node 100 comprises suitable means, such as e.g. the processor 102 and the transceiver 104, configured to perform said actions.

According to embodiments of the invention the first network node 100 is configured to serve a first cell 510a using beamforming and to determine a first set of client devices 600a, 600b,... , 600n served by the first cell 510a, wherein the first set of client devices 600a, 600b,... , 600n have correlated channels (see Fig. 5). The first network node 100 is further configured to determine a channel representation indicating a representation of the correlated channels of the first set of client devices 600a, 600b,... , 600n and transmit the channel representation to a second network node 300.

Fig. 2 shows a flow chart of a corresponding method 200 which may be executed in a first network node 100 configured to serve a first cell 510a, such as the first network node 100 shown in Fig. 1 . The method 200 comprises determining 202 a first set of client devices 600a, 600b,... , 600n served by the first cell 510a, wherein the first set of client devices 600a, 600b,... , 600n have correlated channels. The method 200 further comprises determining 204 a channel representation indicating a representation of the correlated channels of the first set of client devices 600a, 600b,... , 600n and transmitting 206 the channel representation to a second network node 300.

Fig. 3 shows a second network node 300 according to an embodiment of the invention. In the embodiment shown in Fig. 3, the second network node 300 comprises a processor 302, a transceiver 304 and a memory 306. The processor 302 is coupled to the transceiver 304 and the memory 306 by communication means 308 known in the art. The second network node 300 may be configured for both wireless and wired communications in wireless and wired communication systems, respectively. The wireless communication capability is provided with an antenna or antenna array 310 coupled to the transceiver 304, while the wired communication capability is provided with a wired communication interface 312 coupled to the transceiver 304. That the second network node 300 is configured to perform certain actions can in this disclosure be understood to mean that the second network node 300 comprises suitable means, such as e.g. the processor 302 and the transceiver 304, configured to perform said actions.

According to embodiments of the invention the second network node 300 is configured to serve a second cell 510b using beamforming and to receive a channel representation from a first network node 100, wherein the channel representation indicates a representation of correlated channels of a first set of client devices 600a, 600b,... , 600n served by a first cell 510a of the first network node 100 (see Fig. 5). The second network node 300 is further configured to determine a precoder for beamforming towards a second set of client devices 600a', 600b',... , 600n' served by the second cell 510b based on the channel representation.

Fig. 4 shows a flow chart of a corresponding method 400 which may be executed in a second network node 300 configured to serve a second cell 510b using beamforming, such as the second network node 300 shown in Fig. 3. The method 400 comprises receiving 402 a channel representation from a first network node 100, wherein the channel representation indicates a representation of correlated channels of a first set of client devices 600a, 600b, ... , 600n served by a first cell 510a of the first network node 100. The method 400 further comprises determining 404 a precoder for beamforming towards a second set of client devices 600a', 600b',... , 600n' served by the second cell 510b based on the channel representation. Fig. 5 shows a communication system 500 according to an embodiment of the invention. In the embodiment shown in Fig. 5, the communication system 500 comprises a first network node 100, a second network node 300, and a number of client devices 600n configured for wireless communication in the communication system 500. In embodiments, the communication system 500 can be based on NR or LTE radio access technology (RAT) but is not limited thereto.

The first network node 100 serves a first cell 510a and the second network node 300 serves a second cell 510b. The first cell 510a and second cell 520b each implements a number of beams using beamforming to serve the number of client devices 600. Each client device which is served by a cell may be associated with a single or multiple beams within that cell. With reference to Fig. 5, the first cell 510a serves two client devices 600a, 600b with one beam, while the second cell 520b serves three client device 600a’, 600b’, 600c’ with a respective beam. The shown client devices 600n are all cell-edge client devices, i.e. client devices located close to the border of a cell, and may hence experience inter-cell interference. To avoid inter- cell interference for cell-edge client devices CBF can be performed between the first network node 100 and the second network node 300.

According to embodiments of the invention an improved CBF procedure is provided allowing the first network node 100 and the second network node 300 to exchange information which is representative of a cluster of cell-edge client devices 600, herein referred to as a first set of client devices 600a, 600b,... , 600n. The first set of client devices 600a, 600b,... , 600n have radio channels which are correlated and may characterize an inter-cell interference area. In the embodiment shown in Fig. 5, the first set of client devices 600a, 600b,... , 600n comprises two client device 600a, 600b served by the first cell 510a. By exchanging the information according to the invention representing the first set of client devices 600a, 600b, the second network node 300 may determine an inter-cell precoding matrix for its scheduled client devices 600a’, 600b’, 600c’, while taking into account the information about the first set of client devices 600a, 600b from the first network node 100. The second network node 300 may hence determine and use a precoder for beamforming towards its scheduled client devices 600a’, 600b’, 600c’, which avoids an interfering beam towards the client devices 600a, 600b in the first cell 510a, as indicated by the dashed beam in Fig. 5. Thereby, the client devices 600a, 600b in the first cell 510a will receive less interference.

Fig. 6 illustrates coordinated beamforming between the first network node 100 and the second network node 300 according to an embodiment of the invention. The first network node 100 serves the first cell 510a using beamforming and the second network node 300 serves the second cell 510b using beamforming.

In step I in Fig. 6, the first network node 100 determines a first set of client devices 600a, 600b,... , 600n served by the first cell 510a, where the first set of client devices 600a, 600b,... , 600n have correlated channels. The first set of client devices 600a, 600b,... , 600n may be client devices located close to the edge of the first cell 510a and further close to the second cell 510b served by the second network node 300. Thus, the first set of client devices 600a, 600b,... , 600n may be client devices located in an area subject to inter-cell interference from the second cell 510b which may benefit from coordinated beamforming.

The first network node 100 may determine the first set of client devices 600a, 600b,... , 600n based on at least one of a channel correlation distance, a geographical distance or a location. The first network node 100 may e.g. look at the channel correlation distance, the geographical distance and/or the location associated with each client device and determine the first set of client devices 600a, 600b,... , 600n 600a, 600b,... , 600n to be client devices with a channel correlation distance, a geographical distance or a location within a predetermined distance or location, respectively.

In embodiments, the first network node 100 may hence determine the first set of client devices 600a, 600b,... , 600n based on a comparison of the channel correlation distance and a channel correlation distance threshold value. In a similar way, the first network node 100 may further determine the first set of client devices 600a, 600b,... , 600n based on a comparison of the geographical distance and a geographical distance threshold, or based on a comparison of the location and a location threshold.

In step II in Fig. 6, the first network node 100 determines a channel representation indicating a representation of the correlated channels of the first set of client devices 600a, 600b,... , 600n. The first network node 100 may determine the channel representation based on statistics of the correlated channels of the first set of client devices 600a, 600b,... , 600n. The channel representation may e.g. correspond to a correlated channel of the first set of client devices 600a, 600b,... , 600n with an average correlation distance, a minimum correlation distance, a maximum correlation distance, a correlation distance within a high or low correlation distance variance, or may correspond to multiple correlated channels from the correlation distance distribution of the first set of client devices 600a, 600b,... , 600n, or any combination thereof. The first network node 100 may further determine the channel representation based on at least one of: client device mobility, client device traffic activity, client device channel stability and client device interference. The client device mobility may comprise mobility information such as whether a client device is stationary or mobile and/or mobility of the average, slowest and/or fastest client device in the first set of client devices 600a, 600b,... , 600n in terms of velocity and direction. The client device traffic activity may comprise traffic information such as traffic of an average client device, traffic of the client device with the highest and/or lowest traffic demands, and/or aggregated traffic demands of all client devices in the first set of client devices 600a, 600b,... , 600n in terms of a non-empty set of QoS parameters and traffic characteristics. The client device channel stability and client device interference may comprise channel information such as the channel of an average client device, an average stationary client device, and/or an average mobile client device in the first set of client devices 600a, 600b,... , 600n in terms of coherence time or bandwidth and/or channel state related characteristics indicating stability and variability.

Consequently, the channel representation may correspond to a correlated channel of a client device of the first set of client devices 600a, 600b,... , 600n that is stationary or mobile having average, lowest or highest speed and velocity values. The channel representation may further correspond to a correlated channel of a client device of the first set of client devices 600a, 600b,... , 600n with certain channel stability as given by average, minimum, or maximum coherence time and/or bandwidth and channel interference, as given by worst or average interfering channel, or geographically being in the center of the geographical area within the geographical distance. The channel representation may further correspond to a correlated channel of a client device of the first set of client devices 600a, 600b, ... , 600n that has average traffic load, peak traffic load, minimum traffic demands, or traffic of certain quality demands.

The channel representation may indicate an identity of a selected first client device 600n in the first set of client devices 600a, 600b,... , 600n and may further indicate a sequence code, and a frequency pattern and a time pattern for the selected first client device 600n in the first set of client devices 600a, 600b,... , 600n. The identity of the selected first client device 600n, and optimally the sequence code, the frequency pattern and the time pattern for the selected first client device 600n, may be used by the second network node 300 for uplink reference signals and/or measurements, as will be described below.

The channel representation may indicate channel coefficient information which can be used by the second network node for downlink reference signals and/or measurements, as will be described below. In step III in Fig. 6, the first network node 100 transmits the channel representation to the second network node 300. The transmission of the channel representation may be performed by means of a fixed or a wireless transmission between the first 100 and the second 300 network node via direct or backhaul connections e.g. implementing Xn interface (or another radio access network node interface according to 3GPP or IEEE standardization) over a singlelink or multi-link communication protocol such as IP.

The second network node 300 receive the channel representation from a first network node 100 and hence the indication of the representation of correlated channels of the first set of client devices 600a, 600b,... , 600n served by the first cell 510a of the first network node 100. Based on the received channel representation, the second network node 300 determines a precoder for beamforming towards a second set of client devices 600a', 600b',... , 600n' served by the second cell 510b, in step IV in Fig. 6.

When the received channel representation indicates an identity of a selected first client device 600n in the first set of client devices 600a, 600b,... , 600n, the second network node 300 receives reference signals from the selected first client device 600n based on the identity of the selected first client device 600n. When the received channel representation further indicates a sequence code, and a frequency pattern and a time pattern for the selected first client device 600n in the first set of client devices 600a, 600b,... , 600n, the second network node 300 further decodes the reference signals based on the sequence code, the frequency pattern and the time pattern for the selected first client device 600n to obtain channel coefficient information and determine the precoder for beamforming based on the channel coefficient information.

When the received channel representation indicates channel coefficient information, the second network node 300 further determines the precoder for beamforming based on the channel coefficient information.

According to embodiments of the invention the first network node 100 determines a time parameter indicating a valid time period for the channel representation based on at least one of: client device mobility, client device traffic activity, client device channel stability and client device interference and transmit the time parameter to the second network node 300. The channel representation transmitted by the first network node 100 in step III in Fig. 6 may hence in embodiments be associated with a valid time period which is indicated to the second network node 300 by the time parameter. The time parameter may be transmitted in the same message as the channel representation or be transmitted in a separate message.

Upon receiving the time parameter indicating the valid time period for the channel representation from the first network node 100, the second network node 300 performs beamforming in the second cell 510b according to the precoder during the valid time period. In other words, the precoder determined by the second network node 300 in step IV in Fig. 6 may in embodiments only be valid for a limited time period and hence be invalid after the end of the valid time period.

On the other hand, if no time parameter indicating a valid time period for the channel representation is received, the second network node 300 may assumed that the received channel representation is valid until a new channel representation is received from the first network node 100. The first network node 100 may determine and transmit a new channel representation if the first set of client devices 600a, 600b,... , 600n change, e.g. due to client device movement, traffic congestion, users gathering in public areas, etc. The first network node 100 may further determine and transmit a new channel representation at time intervals. The channel representation may e.g. be transmitted periodically, where the periodicity may be estimated as a function of the channel coherence time of the client devices in the first set of client devices 600a, 600b,... , 600n. The channel coherence time may e.g. be an average, a maximum, and/or a minimum channel coherence time, a channel coherence time of the most representative client device, as well as other statistical or arithmetical measure of central tendency and dispersion of channel coherence time for different sizes of confidence intervals. The channel representation may further be transmitted periodically based on fixed periods of certain length or non-periodically based on higher-layer statistics of a channel covariance matrix associated with the first set of client devices 600a, 600b,... , 600n. Low variance may lead to a longer period being used, while high variance may indicate shorter coherence time of the channel representation and thus lead to a shorter period. The variance levels mapping to a period can further be represented by a linear function.

According to embodiments of the invention, the first network node 100 may further determine scheduling information associated with the first set of client devices 600a, 600b,... , 600n and transmit the scheduling information to the second network node 300. The scheduling information indicates if a client device within the first set of client devices 600a, 600b,... , 600n will be scheduled and can be used by the second network node 300 for scheduling coordination. In embodiments, the scheduling information may have a zero value if no client device in the first set of client devices 600a, 600b,... , 600n is to be scheduled and have a positive number value if at least one client device in the first set of client devices 600a, 600b, ... , 600n is to be scheduled. If the second network node 300 receives a channel representation but no scheduling information for the first set of client devices 600a, 600b, ... , 600n, the second network node 300 may assume by default a scheduling information value larger than zero and perform interference coordination.

Further details related to calculating a precoder matrix in the second network node 300 taking the first set of client devices 600a, 600b,... , 600n into account will now be described. A client device in the first set of client devices 600a, 600b,... , 600n, i.e. a client device served by a neighbor cell, may be referred to as a virtual client device to the second network node 300. Adding a virtual stream for virtual client devices in the precoding matrix W of the second network node 300 can be done using eigen-vector zero forcing (EZF) or regularized eigenvector zero forcing (REZF) as follows:

Multi-cell joint EZF w = v^H(VV^H)^-1 where V = [w₁, w₂, ... , w_/, v₁, v₂, ..., v₇], w_t e servcell, and Vj ∈ inter f cell, I is the number of served client devices, J is the number of interfered client devices, w_{i ,} i = 1,2, ..., /, and v_J = 1,2, ...,J, are vectors of channel coefficients of the j-th served and /-th interfered client device respectively.

Multi-cell joint REZF w = v^H(VV^H + D)^-1 where V = [w_1; w₂, ... , w_I, v₁, v₂, ... , v_j , w_i ∈ servcell, and Vj ∈ inter f cell are constructed as for EZF and D is given by

where L is the total number of pairing layers, = SINR_i ∈ servcell , and y'_j = SINR_i ∈ interfcell.

In both EZF and REZF, the elements of the channel coefficient vectors w_t, i = 1,2, ..., /, and vj = 1,2, ...,/, corresponds to the multiple-antenna or beam connections of the j-th served and j-th interfered, respectively.

The inverse matrix in the EZF and REZF precoders may not be invertible in case of high correlated layers. One way to remedy this is to check the inverse matrix (determinant equal to zero) when adding one virtual client device, but it will require a lot of computations. Another way is using e.g. machine learning based or rate estimation algorithm to choose cell-edge client devices before precoding.

A virtual client device could be one specific client device in the first set of client devices 600a, 600b,... , 600n or a statistically representative client device representing an average client device of the first set of client devices 600a, 600b,... , 600n, as shown in Figs. 7a-b.

Figs. 7a-b show embodiments where the first set of client devices 600a, 600b,... , 600n comprises a first client device 600a and a second client device 600b served by the first cell 510a. In Fig. 7a, complete client device information, i.e. information about both the first client device 600a and the second client device 600b, is exchanged. The channel representation transmitted by the first network node 100 to the second network node 300 hence indicates both the channel of the first client device 600a

and the channel of the second client device 600b v₂.

Fig. 7a illustrates a baseline approach where the precoding matrix W of the second network node 300 can be determined using EZF or REZF as follows:

Multi-cell joint EZF baseline approach

In one embodiment when multi-cell joint EZF precoding is used, the precoding matrix W is given by W = V^H(VV^H)^-1 with:

- V = [w_1; w₂, w₃, V₁] , in the occasion that the first network node 100 schedules for transmission the first client device with correlated channel as measured by the second network node 300, wherein the transmission of the first client device uses the same resources as the second set of client devices with correlated channels w₁,w₂,w₃ as measured by the second network node 300.

- V = [w_1; w₂,w₃, v₁], in the occasion that the first network node 100 schedules for transmission the first client device with correlated channel v₂ as measured by the second network node 300, wherein the transmission of the first client device uses the same resources as the second set of client devices with correlated channels w₁,w₂,w₃ as measured by the second network node 300.

- V = [w_1; w₂,w₃, v₁,v₂], in the occasion that the first network node 100 schedules for transmission first client devices with correlated channel

and v as measured by the second network node 300, wherein the transmission of the first client devices uses the same resources as the second set of client devices with correlated channels w₁, w₂, w₃ as measured by the second network node 300.

Multi-cell joint REZF baseline approach

In one embodiment when multi-cell joint EZF precoding is used, the precoding matrix W is given by W = V^H(VV^H + D )^-1 where:

- V = [w_1; w₂,w₃,v₁], D is given by D = L . diag , L is the total number of

pairing layers, y₁ , y₂ , y₃ are the signal-to-interference-and-noise ratio of the client devices in the second set with correlated channels w₁, w₂, w₃, and y₁, is the signal-to- interference-and-noise ratio of the first client device with correlated channel v₁ in the occasion that the first network node schedules for transmission the first client device with correlated channel as measured by the second network node 300, wherein the transmission of the first client device uses the same resources as the second set of client devices with correlated channels w₁, w₂, w₃ as measured by the second network node 300.

- V = [w_1; w₂, w₃, v₂], D is given by D = L . diag L is the total number of

pairing layers, y₁ , y₂ , y₃ are the signal-to-interference-and-noise ratio of the client devices in the second set with correlated channels w₁, w₂, w₃, and y'₂, is the signal-to- interference-and-noise ratio of the first client device with correlated channel v₂, in the occasion that the first network node 100 schedules for transmission the first client device with correlated channel v₂ as measured by the second network node 300, wherein the transmission of the first client device uses the same resources as the second set of client devices with correlated channels w₁, w₂, w₃ as measured by the second network node 300.

- V = [w_1; w₂, w₃, v_lr v₂] , D is given by D = L . diag , L is the total

number of pairing layers, y₁, y₂, y₃ are the signal-to-interference-and-noise ratio of the client devices in the second set with correlated channels w_1; w₂, w₃, and y₁, and y'₂ are the signal-to-interference-and-noise ratio of the first client devices with correlated channel and v₂ , in the occasion that the first network node 100 schedules for transmission the first client devices with correlated channel and v as measured by the second network node 300, wherein the transmission of the first client devices uses the same resources as the second set of client devices with correlated channels w₁, w₂, w₃ as measured by the second network node 300. A drawback of the above reference approaches may be that it requires the first network node 100 to exchange information at any scheduling time about the channel representation of all client devices it is about to schedule. Such a solution is impractical for two reasons: (a) it is not a scalable solution as it requires large amount of channel representation to be exchanged, (b) it is not robust solution since the exchange delay significantly exceeds the scheduling time intervals, that is, the time between scheduling decisions, turning the channel representation information obsolete.

In Fig. 7b, virtual client device information is exchanged. The channel representation transmitted by the first network node 100 to the second network node 300 hence indicates the channel of a virtual client device v₁₂ representing the first set of client devices 600a, 600b,... , 600n. The channel of the virtual client device, v₁₂ may characterize any client device of the first set of client devices 600a, 600b,... , 600n, i.e. either the first client device 600a or the second client device 600b in this embodiment.

Fig. 7b illustrates an embodiment of the invention where the precoding matrix W of the second network node 300 can be determined using EZF or REZF as follows:

Multi-cell joint EZF approach

In an embodiment when multi-cell joint EZF precoding is used, the precoding matrix W is given by W = V^H(VV^H)^_1 with:

- V = [w_1; w₂, w₃, v₁₂], no matter if the first network node 100 schedules for transmission the first client device with correlated channel or first client device with correlated channel v₂ or both simultaneously. The correlated channel v₁₂ of the so called virtual client device, is based on the client devices of the first set as determined by the first network node 100 v₁₂ = f (v_1; v₂), and the transmission of the first client device uses the same resources as the second set of client devices with correlated channels w₁, w₂, w₃ as measured by the second network node. The correlated channel of the so called virtual client device, v₁₂ , which represents the first set of client devices with correlated channels V₁ and v₂ , can be derived as a function f of and v₂ , i.e. v₁₂ = f(v₁, v₂) which is further detailed below.

Multi-cell joint REZF approach

In an embodiment when multi-cell joint EZF precoding is used, the precoding matrix W is given by W = V^H(VV^H + D )^-1 where: - V = [w_1; w₂, w₃, v₁₂], D is given by D = L . diag , L is the total number of

pairing layers, γ₁ , y₂ , y₃ are the signal-to-interference-and-noise ratio of the client devices in the second set with correlated channels w_1; w₂, w₃, and y'₁₂, is the signal-to- interference-and-noise ratio of the so called virtual client device, which is based on the client devices of the first set as determined by the first network node 100 v₁₂ = f (v_1; v₂), and used no matter if the first network node 100 schedules for transmission the first client device with correlated channel or the first client device with correlated channel v or both simultaneously, and wherein the transmission of the first client devices uses the same resources as the second set of client devices with correlated channels w₁, w₂, w₃ as measured by the second network node. The correlated channel of the so called virtual client device, v₁₂ , which represents the first set of client devices with correlated channels Vi and v₂ , can be derived as a function f of and v₂ , i.e. v₁₂ = f(v₁, v₂) which is further detailed below.

The benefit of exchanging the channel representation of one virtual client device instead of every client device in the first set of client devices is beneficial in terms of scalability. To improve robustness the channel presentations of more than one virtual client device can be exchanged assuming that the so called virtual client devices of the first set of client devices effectively represents the correlated channel distribution of the first set of client devices.

The channel of the virtual client device v₁₂ can be derived as a function f of and v₂, i.e. ^v12 = This function is the description of the virtual client device, i.e. a representation of the correlated channels of the first set of client devices 600a, 600b,... , 600n, and may be constructed based on:

• the most representative client device of the potentially scheduled client devices, v_{1 ,} v₂, in the first set of client devices 600a, 600b,... , 600n (one specific selected client device v_j, i = 1,2, e.g. the client device with the highest channel correlations, or the client device with most stable channel,

• the average virtual client device in the first set of client devices 600a, 600b,... , 600n derived based on the element-wise average, f = mean(v₁, v₂, of the potentially scheduled client devices, v_{1 t} v₂, in the first set of client devices 600a, 600b,... , 600n, e.g. the client device that is closer to the average channel of all client devices in the first set of client devices 600a, 600b,... , 600n,

• the minimum virtual client device in the first set of client devices 600a, 600b,... , 600n derived based on the element-wise minimum, f = min v₁, v₂, of the potentially scheduled client devices, v₁ v₂, in the first set of client devices 600a, 600b,... , 600n, e.g. the client device with the lowest signal-to-interference-plus-noise ratio (SINR) among the client devices of the first set of client devices 600a, 600b,... , 600n,

• the maximum virtual client device in the first set of client devices 600a, 600b,... , 600n derived based on the element-wise maximum, f = max(v₁,v₂, of the potentially scheduled client devices, v_{1 ,} v₂, in the first set of client devices 600a, 600b,... , 600n, e.g. the client device with the highest SINR among the client devices of the first set of client devices 600a, 600b,... , 600n,

• the average and the variance virtual client device in the first set of client devices 600a, 600b,... , 600n derived based on the element-wise average, f = mean(v₁, v₂, -. vj), and the element-wise variance, f = var v₁, v₂, of the potentially scheduled client devices, v_{1 ,} v₂, in the first set of client devices 600a, 600b,... , 600n, e.g. the client device whose channel and variance is closer to the average channel of all client devices in the first set of client devices 600a, 600b,... , 600n,

• the virtual client device in the first set of client devices 600a, 600b,... , 600n derived based on any other related element-wise algebraic, arithmetic, statistical, similarity, interpolation, correlation or state description operation on channel, traffic, and mobility characteristics of the potentially scheduled client devices, v_{1 ,} v₂, in the first set of client devices 600a, 600b,... , 600n, and

• alternatively a set of representative virtual client devices selected so as to closely or approximately exhibit the average and the variance of the potentially scheduled client devices, v_{1 ,} v₂, in the first set of client devices 600a, 600b,... , 600n.

When selecting the most representative client device in the first set of client devices 600a, 600b,... , 600n as the virtual client device, the selection choice can be further based on the mobility considerations since selecting a client device that does not move fast and/or much may represent the first set of client devices 600a, 600b,... , 600n better. Furthermore, when selecting the most representative client device in the first set of client devices 600a, 600b,... , 600n as the virtual client device, the selection choice can be further based on traffic of client devices. Selecting a client device that has higher traffic demands, or traffic demands that are close the traffic demands of the average client device in the first set of client devices 600a, 600b,... , 600n, may represent the first set of client devices 600a, 600b,... , 600n better. Depending on the first set of client devices 600a, 600b,... , 600n the traffic demands could be determined in terms of bit rate, delay, latency, packet or frame or block loss, or other traffic related QoS parameters. The virtual client device can further be represented by the client device that creates highest interference to any neighbor cell client device, or the most centered client device in the first set of client devices 600a, 600b, ... , 600n, or the client device with traffic volume which is close to the average client device traffic volumes in the first set of client devices 600a, 600b,... , 600n or the client device with a high or even the highest traffic volume, if very uneven traffic usage is observed in the first set of client devices 600a, 600b,... , 600n.

Providing such compressed statistical information for the channel of the virtual client device v₁₂ and as calculated over a time interval τ would allow for the joint scheduling and precoding of any client device in the corresponding area of the first set of client devices 600a, 600b,... , 600n, see dashed circle in Fig. 7b, for that time interval τ. This information from a neighbor cell is passed to L1 to effectively perform distributed multi-cell precoding EZF or REZF. In this case, the receiving cell may determine how to process and include v₁₂ in the precoding matrix. The receiving cell may generate one or multiple virtual client devices according to the statistics, i.e. variance of the client devices of the first set of client devices 600a, 600b,... , 600n. The advantage of such an approach is that instead of exchanging information about every client device the signaling between cells is reduced to the minimum of one virtual client device related information.

Depending on the division duplex technique used, different embodiments can be envisaged for the virtual client device precoding information. The exchange of the information provides the information needed at the receiving end to extract or generate the channel coefficients of the virtual client device v_j, i = 1,2, ...,J.

In frequency division duplexing (FDD), relevant information, e.g. channel state information (CSI) or uplink sounding reference signal (SRS), may be prepared for each first set of client devices 600a, 600b,... , 600n by the serving cell and send to the interfered cell, e.g.:

• Average CSI of all client devices of the first set of client devices 600a, 600b,... , 600n,

• Worst CSI of all client devices of the first set of client devices 600a, 600b,... , 600n,

• Best CSI of all client devices of the first set of client devices 600a, 600b,... , 600n.

In time division duplexing (TDD), cells can receive the uplink reference signal (UL RS) of the first set of client devices 600a, 600b,... , 600n connected to neighboring cells. To decode the signal the following information may be exchanged:

• the sequence code and the SRS allocation pattern in frequency and time of a client device that would be representative of the client devices within the first set of client devices 600a, 600b,... , 600n, • the sequence code and the SRS allocation pattern in frequency and time of the worst client device within the first set of client devices 600a, 600b,... , 600n,

• the sequence code and the SRS allocation pattern in frequency and time of the best client device within the first set of client devices 600a, 600b,... , 600n.

The statistical description of the virtual client device can be stored in a database through statistical processing and be accessible by a cell.

As previously described, the first set of client devices 600a, 600b,... , 600n can be defined based on their inter-correlation values. It may be expected that highly correlated client devices of an interfering cell, as identified over the target bandwidth, belong to the same first set of client devices 600a, 600b,... , 600n. It may further be expected that the client devices within such a first set of client devices 600a, 600b,... , 600n coincide spatially and within certain statistical confidence. In a similar manner, highly correlated client devices of a serving cell may belong to the same first set of client devices 600a, 600b,... , 600n. Clustering client devices in such a way is advantageous since pairing based on client device correlation can now be performed on a client device cluster basis. Pairing implies that client devices with low correlation can be assigned on the same time and frequency resources allowing for efficient spatial multiplexing. By picking one client device from each first set of client devices 600a, 600b,... , 600n with average correlation below a threshold, spatial multiplexing groups can be created. The advantage is that the number of correlation evaluations is decreased assuming optimal correlation threshold. Too conservative thresholds result in a large number of first sets of client devices 600a, 600b,... , 600n, whereas more opportunistic thresholds result in large and volatile first sets of client devices 600a, 600b,... , 600n. Optimal correlation threshold may further differ from location to location.

The same strategies and statistical descriptions associated to beams can be performed at grid resolution. In this case, the first network node 100 and the second network node 300 may exchange information indicating virtual client device precoder information for each grid having cell-edge client devices. Correlation coefficients between grids can be derived statistically and stored in a radio service map (RSM) database. The statistical description of the virtual client device can be stored in RSM through statistical processing and be accessible by a cell. Based on this information a cell is able to coordinate L1 precoding with L2 scheduling depending on the correlation of the interference beam vector and the scheduled beam vector pairs in a grid. In a grid implementation, the grid resolution can be used as an optimization variable, i.e. a fixed grid resolution can be used. Alternatively, different locations may require different grid sizes depending on the variation of correlation, i.e. a flexible grid resolution can be used.

The above channel representation of the so called virtual user and its valid time period information exchange can be extended to include the mobility, the traffic, and the channel characteristics of the virtual user.

In particular, the exchanged information channel presentation and valid time period can be extended to include traffic information comprising: the traffic of the average client device in the first set of client devices comprising a nonempty set of QoS parameters and traffic characteristics of the average client device in the first set of client devices, or the traffic of the client device with the highest/lowest traffic demands in the first set of client devices comprising a non-empty set of QoS parameters and traffic characteristics, or the aggregate traffic demands of all client devices in the first set of client devices comprising a non-empty set of QoS parameters and traffic characteristics, or any combinations thereof.

Furthermore, the exchanged information channel presentation and valid time period can be extended to include mobility information comprising: the mobility of the average client device in the first set of client devices comprising velocity and direction, or the mobility of the fastest/slowest client device in the first set of client devices comprising velocity and direction, or any combinations thereof.

Additionally, the exchanged information channel presentation and valid time period can be extended to include channel information comprising: the channel of the average client device in the first set of client devices comprising coherence time or bandwidth and/or channel state related characteristics indicating stability and variability, or the channel of the average stationary client device in the first set of client devices comprising coherence time or bandwidth and/or channel state related characteristics indicating stability and variability, or the channel of the average mobile client device in the first set of client devices comprising coherence time or bandwidth and/or channel state related characteristics indicating stability and variability, or any combinations thereof.

In addition, the exchanged information channel presentation and valid time period can be extended to include any combination of the traffic, mobility and channel information listed above.

The determination of the channel representation of the so called virtual user can be based on any combination of mobility, the traffic, and the channel characteristics as suggested above. More specifically, with regards to the traffic characteristics the most representative user to selected for the derivation of the virtual user, v₁₂, may comprise: the client device whose traffic is closer to the traffic of average client device in the first set of client devices comprising a non-empty set of QoS parameters and traffic characteristics, or the client device whose traffic has the highest/lowest traffic demands in the first set of client devices comprising a non-empty set of QoS parameters and traffic characteristics, or the client device whose mobility is closer to the mobility of the average client device in the first set of client devices comprising velocity and direction, or the client device whose mobility is that of the fastest/slowest client device in the first set of client devices comprising of velocity and direction, or the client device whose channel is closer to the channel of the average user client device in the first set of client devices comprising coherence time or bandwidth and/or channel state related characteristics indicating stability and variability, or the client device whose channel is closer to the channel of the average stationary client device in the first set of client devices comprising coherence time or bandwidth and/or channel state related characteristics indicating stability and variability, or the client device whose channel is closer to the channel of the average mobile client device in the first set of client devices comprising coherence time or bandwidth and/or channel state related characteristics indicating stability and variability, or any combinations thereof.

CBF according to embodiments of the invention provides gains in terms of cell and cell-edge throughput as compared to a scenario without coordination between cells. A performance evaluation of the CBF according to the invention has been performed by means of simulations. In the simulation experiment, a comparison of coordinated scheduling among multiple cells according to the invention is compared to an approach where multiple cells do not coordinate their scheduling at L2.

A wireless system that operates at 3.5GHz and a radio network topology that consists of one site comprising 3 sectors, each corresponding to a cell, is assumed. To create an interfering scenario, the cells are positioned with an inter-site distance of 300m with their antenna array systems targeting towards the center of the simulated area. Each cells antenna array system consists of 128 antenna elements (8x8x2) to create 64 transmitter ports or beams in the downlink. The corresponding receiver ports at client devices, here called UEs, which has a rank of 2, are 4. A rank of 2 in this simulation scenario means that for each client device the set of serving beams can be up to two. Similarly, the set of strongest interfering beams may consist by up to two different beams, 1st and 2nd strongest interfering beam. Within each cell 15 client devices are dropped, all having a speed of 3km/h and full buffer traffic. The simulation parameters and assumptions of this study are summarized in Table I.

Table I: simulation parameters and assumptions. For the simulations two schemes have been implemented and evaluated: (1) centralized CBF, where joint L2 scheduling and L1 coordination is performed centrally, and (2) distributed CBF, where joint L2 scheduling and L1 coordination is performed in a distributed manner among the cells.

In the centralized CBF, all information is directly accessed from a common memory including (a) client device scheduling information of neighboring cells, (b) CBF client devices of the neighboring cells (interfered client devices), and (c) the beam each client device is associated with. All client devices are ranked per transmission time interval (TTI) in a priority list according to their proportional fair (PF) score. For each RBG scheduling occasion next high priority client device from the priority list is scheduled in a way that the interference it causes to or receives by neighboring cells’ beams is checked in terms of correlation between i. the serving beams of serving cell and the interference avoidance area, that is the region, which is identified by scheduled CBF client devices of the neighboring cell ii. the interference measurement of CBF client devices by neighboring cell and the accumulated scheduled beam region of the neighboring cell

A client device is scheduled only if the correlations of its interference level to and from neighboring cells are below certain values. Beam regions need to be updated after a client device scheduling pairing is completed, like the interference avoidance area in neighboring cells and the accumulated scheduling areas in the serving cells. Client devices are ranked based on their instantaneous PF values and selected according to their PF score. Client devices that have a rank of 2 can be served by maximum two beams selected based on beam selection threshold of -10dB.

In the distributed CBF scheme, each site/cell ranks its client devices independently and maintains own PF priority lists. To each beam a partial L1 precoding virtual client device information is exchanged between the cells. No explicit scheduling information is exchanged between the cells.

The precoding information is that of a representative virtual client device, corresponding to the worst client device, i.e. client device with minimum SINR. In particular, the following joint L2 coordination and L1 precoding schemes have been simulated:

• NOCS_SCP_REZF: Regularized enhanced zero-forcing (REZF) without joint multi-cell pre-coding at L1 and no coordinated scheduling (baseline approach)

• CS_SCP_REZF: Coordinated scheduling at L2 with single cell pre-coding (SCP) at L1 , i.e. without joint multi-cell pre-coding (REZF) • CS_MCP_REZF: Coordinated scheduling at L2 with joint multi-cell pre-coding (REZF) at L1

• CS_MCP_REZF_CHECK: Joint pre-coding with REZF (with invertible checking for each virtual client device) at L1 , and coordinated scheduling at L2.

It has to be noted that coordinated scheduling in centralized CBF is performed by means of central client device scheduling for all cells simultaneously while in distributed CBF the scheduling is performed by each cell separately by means of L2 coordination.

The beam correlation threshold value for L2, BeamCheckThreshold, is set to 0.3, while the L1 REZF correlation threshold, EZFCheckThreshold, is set to 0.1. It is further assumed that there is a signaling delay between any two cells equal to 5ms.

Figs. 8a-b show performance of the centralized CBF solution based on virtual client device information exchange as compared with the no coordination scenario and the distributed CBF solution with a CBF threshold of 9dB. Fig. 8a shows the cell throughput and Fig. 8b shows the cell-edge throughput. In the distributed scheme information about virtual CBF client devices is exchanged every 5 TTIs with 5 TTIs signaling delay with the virtual client device information exchange corresponding to the worst case client device, i.e. the client device with the minimum SINR.

Compared to the case of no coordination NOCS_SCP_REZF the distributed CBF scheme CS_MCP_REZF shows gains in the order of 7.7% and 51% in cell throughput and cell-edge throughput respectively. The losses of the distributed CBF scheme as compared to the centralized CBF is approx. 9.5% and 14% in terms of cell throughput and cell-edge throughput when CS_MCP_REZF is used. Again, the losses in cell throughput are lower when invertibility checking is applied while the cell edge is slightly improved. In particular, the cell throughput and cell-edge throughput when CS_MCP_REZF_CHECK is used is approx. 7.5% and 1.9% respectively. This is attributed to the fact that the worst case client device is used for the nulling. Overall embodiments of the invention performs significantly better than a solution with no coordination, and retain cell-edge user throughput gains as compared to a centralized solution that ideally have full knowledge about all client devices but is realistically not feasible.

The first network node 100 and the second network node 300 in this disclosure includes but is not limited to: a NodeB in wideband code division multiple access (WCDMA) system, an evolutional Node B (eNB) or an evolved NodeB (eNodeB) in LTE systems, or a relay node or an access port, or an in-vehicle device, a wearable device, or a gNB in the fifth generation (5G) networks. Further, the first network node 300 herein may be denoted as a radio first network node, an access first network node, an access port, or a base station, e.g. a radio base station (RBS), which in some networks may be referred to as transmitter, “gNB”, “gNodeB”, “eNB”, “eNodeB”, “NodeB” or “B node”, depending on the technology and terminology used. The radio first network nodes may be of different classes such as e.g. macro eNodeB, home eNodeB or pico base station, based on transmission power and thereby also cell size. The radio first network node can be a station (STA), which is any device that contains an IEEE 802.11 -conformant MAC and PHY interface to the wireless medium. The radio first network node may also be a base station corresponding to the 5G wireless systems.

The client device 600 in this disclosure includes but is not limited to: a UE such as a smart phone, a cellular phone, a cordless phone, a session initiation protocol (SIP) phone, a wireless local loop (WLL) station, a personal digital assistant (PDA), a handheld device having a communication function, a computing device or another processing device connected to a wireless modem, an in-vehicle device, a wearable device, an integrated access and backhaul node (IAB) such as mobile car or equipment installed in a car, a drone, a device-to-device (D2D) device, a wireless camera, a mobile station, an access terminal, an client device unit, a communication device, a station of wireless local access network (WLAN), a wireless enabled tablet computer, a laptop-embedded equipment, an universal serial bus (USB) dongle, a wireless customer-premises equipment (CPE), and/or a chipset. In an Internet of things (IOT) scenario, the client device 600 may represent a machine or another device or chipset which performs communication with another wireless device and/or a network equipment.

The UE may further be referred to as a mobile telephone, a cellular telephone, a computer tablet or laptop with wireless capability. The UE in this context may e.g. be portable, pocket- storable, hand-held, computer-comprised, or vehicle-mounted mobile device, enabled to communicate voice and/or data, via the radio access network, with another entity, such as another receiver or a server. The UE can be a station (STA), which is any device that contains an IEEE 802.11 -conformant media access control (MAC) and physical layer (PHY) interface to the wireless medium (WM). The UE may also be configured for communication in 3GPP related LTE and LTE-Advanced, in WiMAX and its evolution, and in fifth generation wireless technologies, such as NR.

Furthermore, any method according to embodiments of the invention may be implemented in a computer program, having code means, which when run by processing means causes the processing means to execute the steps of the method. The computer program is included in a computer readable medium of a computer program product. The computer readable medium may comprise essentially any memory, such as a ROM (Read-Only Memory), a PROM (Programmable Read-Only Memory), an EPROM (Erasable PROM), a Flash memory, an EEPROM (Electrically Erasable PROM), or a hard disk drive.

Moreover, it is realized by the skilled person that embodiments of the first network node 100 and the second network node 300 comprise the necessary communication capabilities in the form of e.g., functions, means, units, elements, etc., for performing the solution. Embodiments of other such means, units, elements and functions are: processors, memory, buffers, control logic, encoders, decoders, rate matchers, de-rate matchers, mapping units, multipliers, decision units, selecting units, switches, interleavers, de-interleavers, modulators, demodulators, inputs, outputs, antennas, amplifiers, receiver units, transmitter units, DSPs, MSDs, TCM encoder, TCM decoder, power supply units, power feeders, communication interfaces, communication protocols, etc. which are suitably arranged together for performing the solution.

Especially, the processor(s) of the first network node 100 and the second network node 300 comprise, e.g., one or more instances of a Central Processing Unit (CPU), a processing unit, a processing circuit, a processor, an Application Specific Integrated Circuit (ASIC), a microprocessor, or other processing logic that may interpret and execute instructions. The expression “processor” may thus represent a processing circuitry comprising a plurality of processing circuits, such as, e.g., any, some or all of the ones mentioned above. The processing circuitry may further perform data processing functions for inputting, outputting, and processing of data comprising data buffering and device control functions, such as call processing control, client device interface control, or the like.

Finally, it should be understood that the invention is not limited to the embodiments described above, but also relates to and incorporates all embodiments within the scope of the appended independent claims.

Claims

1. A first network node (100) for a communication system (500), the first network node (100) being configured to serve a first cell (510a) using beamforming, and further configured to determine a first set of client devices (600a, 600b,... , 600n) served by the first cell (510a), wherein the first set of client devices (600a, 600b,... , 600n) have correlated channels; determine a channel representation indicating a representation of the correlated channels of the first set of client devices (600a, 600b,... , 600n); and transmit the channel representation to a second network node (300).

2. The first network node (100) according to claim 1 , wherein determine the first set of client devices (600a, 600b,... , 600n) comprises determine the first set of client devices (600a, 600b,... , 600n) based on a channel correlation distance.

3. The first network node (100) according to claim 2, wherein determine the first set of client devices (600a, 600b,... , 600n) further comprises determine the first set of client devices (600a, 600b,... , 600n) based on a comparison of the channel correlation distance and a channel correlation distance threshold value.

4. The first network node (100) according to any one of the preceding claims, wherein determine the channel representation comprises determine the channel representation based on statistics of the correlated channels of the first set of client devices (600a, 600b,... , 600n).

5. The first network node (100) according to any one of the preceding claims, wherein determine the channel representation comprises determine the channel representation based on at least one of: client device mobility, client device traffic activity, client device channel stability and client device interference.

6. The first network node (100) according to any one of the preceding claims, further configured to determine a time parameter indicating a valid time period for the channel representation based on at least one of: client device mobility, client device traffic activity, client device channel stability and client device interference; and transmit the time parameter to the second network node (300).

7. The first network node (100) according to any one of claims 1 to 6, wherein the channel representation indicates an identity of a selected first client device (600n) in the first set of client devices (600a, 600b,... , 600n).

8. The first network node (100) according to claim 7, wherein the channel representation further indicates a sequence code for the selected first client device (600n) in the first set of client devices (600a, 600b,... , 600n).

9. The first network node (100) according to any one of claims 1 to 6, wherein the channel representation indicates channel coefficient information.

10. A second network node (300) for a communication system (500), the second network node (300) being configured to serve a second cell (510b) using beamforming, and further configured to receive a channel representation from a first network node (100), wherein the channel representation indicates a representation of correlated channels of a first set of client devices (600a, 600b,... , 600n) served by a first cell (510a) of the first network node (100); and determine a precoder for beamforming towards a second set of client devices (600a', 600b',... , 600n') served by the second cell (510b) based on the channel representation.

11. The second network node (300) according to claim 10, further configured to receive a time parameter from the first network node (100), wherein the time parameter indicates a valid time period for the channel representation; and perform beamforming in the second cell (510b) according to the precoder during the valid time period.

12. The second network node (300) according to claim 10 or 11 , wherein the channel representation indicates an identity of a selected first client device (600n) in the first set of client devices (600a, 600b,... , 600n), and wherein the second network node (300) is further configured to receive reference signals from the selected first client device (600n) based on the identity of the selected first client device (600n).

13. The second network node (300) according to claim 12, wherein the channel representation further indicates a sequence code for the selected first client device (600n) in the first set of client devices (600a, 600b,... , 600n), and wherein the second network node (300) is further configured to decode the reference signals based on the sequence code for the selected first client device (600n) to obtain channel coefficient information; and determine the precoder for beamforming based on the channel coefficient information.

14. The second network node (300) according to claim 10 or 11 , wherein the channel representation indicates channel coefficient information, and wherein the second network node (300) is further configured to determine the precoder for beamforming based on the channel coefficient information.

15. A method (200) for a first network node (100) configured to serve a first cell (510a) using beamforming, the method (200) comprising determining (202) a first set of client devices (600a, 600b,... , 600n) served by the first cell (510a), wherein the first set of client devices (600a, 600b,... , 600n) have correlated channels; determining (204) a channel representation indicating a representation of the correlated channels of the first set of client devices (600a, 600b,... , 600n); and transmitting (206) the channel representation to a second network node (300).

16. A method (400) for a second network node (300) configured to serve a second cell (510b) using beamforming, the method (400) comprising receiving (402) a channel representation from a first network node (100), wherein the channel representation indicates a representation of correlated channels of a first set of client devices (600a, 600b,... , 600n) served by a first cell (510a) of the first network node (100); and determining (404) a precoder for beamforming towards a second set of client devices (600a', 600b',... , 600n') served by the second cell (510b) based on the channel representation.

17. A computer program with a program code for performing a method according to claim 15 or 16 when the computer program runs on a computer.