Systems and methods for uplink coordinated transmission in open radio access networks

Abstract

A communication system (100) with an open radio access network architecture is provided. The communication system includes multiple (K) user equipments (3) configured to exchange messages through the communication system. The communication system includes one or more radio units (5) and a distributed unit (6), each radio unit (5) being connected to the distributed unit via a fronthaul link (4) with a given capacity. The communication system includes a coordination system (20) configured to coordinate the exchange between the radio units and the distributed unit (6) in response to signals received by the radio units (5) from the user equipments (3). The coordination system (20) is configured to send signaling messages including data stream selection information to each radio unit (5). Each radio unit includes: - a transformation unit (50) configured to apply a transformation operation to a signal vector received by the radio unit (5), the transformation operation providing a transformed vector, the transformation operation being defined as reducing the dimension of the received signal vector; - a data stream selection unit (52) configured to select multiple data streams corresponding to components of the transformed vector using data stream selection information included in the signaling message; - a quantizer configured to apply a quantization operation to the selected data streams, the quantization operation providing a quantized vector, the radio unit (5) being configured to transmit the quantized vector to the distributed unit (6). On all radio units of the communication system (100), the sum of the number of data streams selected at the radio unit (5) is greater than or equal to the product of the number of user equipments (K) and the number of antennas of the user equipments.

Description

Technical Field

[0001] The present invention relates generally to communication systems, and more particularly to communication systems implemented according to Open Radio Access Network (O-RAN) and coordination methods implemented in such communication systems. Background Technology

[0002] With the widespread use of advanced wireless devices and video streaming, smaller cellular networks have emerged to meet the exponentially growing demand for data rates. As a result, the distance between base stations has become smaller, which in turn leads to greater inter-cell interference.

[0003] Cellular radio access networks (RANs) have traditionally used proprietary network equipment from a small group of network equipment manufacturers. In recent years, the open RAN (also known as O-RAN) architecture has been adopted for 5G standards.

[0004] The O-RAN architecture introduces three distinct components:

[0005] - Radio Unit O-RU (Radio Head),

[0006] - Distributed Unit (O-DU) and

[0007] - Centralized Unit (O-CU).

[0008] An Open Radio Access Network (O-RAN) was developed to address major inter-cell interference and improve network performance by implementing joint coding (downlink DL) and decoding (uplink UL) at the O-DU. In the uplink (UL) scenario, in order to forward the received signal at the O-RU, each O-RU needs to first compress / quantize the signal and then transmit it to the O-DU in a distributed manner via the fronthaul link.

[0009] Two encoding schemes were proposed:

[0010] In D.S. Lepian and J.K. Wolf, “Noiseless coding of correlated information sources”, IEEE Trans. Inf. Theory, vol. 19, pp. 471-480, Jul. 1973, coding as information theory lossless compression over distributed compression is proven to be optimal.

[0011] - In ADWyner and J.Ziv, “The rate-distortion function for source coding with side information at the decoder”, IEEE Trans.Inf.Theory, vol.22, pp.1-10, Jan.1976, a coding scheme for handling lossy distributed compression problems was proposed.

[0012] Both of these encoding schemes are based on shifting computational complexity from the encoder to the decoder. This is suitable for O-RAN systems (cloud RAN) because the computational burden is shifted from the O-RU to the O-DU, while in O-RAN, most of the computation should be at the O-DU.

[0013] However, implementing the two schemes proposed by these technologies may be prohibitively expensive or even impossible. In fact, both schemes are based on rate distortion theory, which assumes an infinite block length for the source coding, an assumption that cannot be verified in real-world systems. Furthermore, the computational complexity of obtaining these two schemes is prohibitively high, especially when the O-RU is equipped with a large number of antennas, due to the need to solve non-convex optimization problems.

[0014] Therefore, there is a need to provide an improved communication system and method in open radio access networks that enables the transmission of as much useful information as possible from the radio unit to the distributed unit, while enhancing the joint decoding performance at the distributed unit. Summary of the Invention

[0015] To address these and other issues, a communication system is provided having an open radio access network architecture. The communication system includes multiple user equipments (User Equipments) configured to exchange messages through the communication system. Each User Equipment includes multiple antennas. The communication system includes one or more radio units and distributed units. Each radio unit is connected to the distributed unit via a fronthaul link with a given capacity. The communication system includes a coordination system configured to coordinate the exchange between the radio units and the distributed units in response to messages received by the radio units from the User Equipments. The messages received at each radio unit are represented by signal vectors. The coordination system is configured to send signaling messages including data stream selection information to each radio unit. Each radio unit includes:

[0016] - A transformation unit configured to apply a transformation operation to a signal vector received by a radio unit, the transformation operation providing a transformed vector, the transformation operation being defined as reducing the dimension of the received signal vector;

[0017] - A data stream selection unit, configured to use the data stream selection information included in the signaling message to select multiple data streams corresponding to the components of the transformed vector.

[0018] - A quantizer configured to apply a quantization operation to a selected data stream, the quantization operation providing a quantized vector.

[0019] The radio unit is configured to send the quantized vector to the distributed unit.

[0020] Wherein, on all radio units of the communication system, the sum of the number of data streams selected at each radio unit is greater than or equal to the product of the number of user equipment and the number of antennas of the user equipment.

[0021] In one embodiment, the distributed unit includes a reconstruction unit and a decoder, the reconstruction unit being configured to perform reconstruction of a received signal vector based on the quantized vector received from all the radio units, the reconstruction providing a reconstructed signal vector, and the decoder being configured to decode the reconstructed signal vector to determine an estimate of a signal transmitted by the user equipment.

[0022] In some implementations, data stream selection information received by a given radio unit via signaling messages may include a selection vector associated with the radio unit, the selection vector having a length corresponding to the number of receiving antennas at the radio unit. A data stream selection unit for a given radio unit may be configured to extract the selection vector corresponding to the given radio unit from the signaling messages and perform a matrix multiplication between the extracted selection vector and the transformed vector, the matrix multiplication providing a vector of the selected data stream, the quantizer of the given radio unit being applied to the vector of the selected data stream.

[0023] The selection vector associated with each radio unit can be determined based on channel state information.

[0024] In one implementation, if the i-th component of the transformed vector must be transmitted from the given radio unit to the distributed unit, the i-th coefficient of the selected vector associated with the given radio unit may have a value equal to 1; conversely, if the i-th component of the transformed vector will not be transmitted from the given radio unit to the distributed unit, it may have a value equal to 0.

[0025] Each selection vector can be determined based on the eigenvalues ​​of the covariance channel matrix.

[0026] A set of selection vectors (V1,...V1) associated with all radio units n ,...V N ) can be determined such that:

[0027]

[0028] Wherein, N represents the number of radio units, and

[0029] V n (i) represents vector V n The i-th element.

[0030] In one implementation, the selection vector can be determined by the distributed unit.

[0031] The selection vector associated with the radio unit may depend solely on statistical information of the channel matrix.

[0032] Specifically, if the k-th and m-th entries of the covariance matrix of the transformed vector have the strongest average power across all channel implementations, then each selected vector V j It can be defined such that the components V of the selection vector j (k) equals the value 1 and the component V of the selection vector j (m) equals 1.

[0033] Alternatively, the selection vector associated with the radio unit may depend solely on the instantaneous implementation of the channel matrix.

[0034] Specifically, the selection vector associated with the radio unit can be determined based on the received signal-to-noise ratio priority.

[0035] In this implementation, the coordination system can be configured to send auxiliary signaling messages between the radio units to exchange priority indices.

[0036] In some implementations, the selection vector associated with the radio unit may be determined based on the condition number priority (CNP) of the channel matrix.

[0037] In one implementation, the coordination system can be distributed, with signaling messages exchanged between radio units.

[0038] A coordination method is also provided for implementation in a communication system having an open radio access network architecture. The communication system includes multiple user equipments exchanging messages through the communication system, which includes one or more radio units and distributed units, each radio unit being connected to the distributed unit via a fronthaul link of given capacity. The coordination method includes coordinating the exchange between the radio unit and the distributed unit in response to the radio unit receiving a message sent from a user equipment, wherein the message received at each radio unit is represented by a signal vector. The method includes sending a signaling message including data stream selection information to each radio unit. The method also includes, at each radio unit:

[0039] - A transformation operation is applied to the signal vector received by the radio unit, the transformation operation providing a transformed vector, the transformation operation being defined as reducing the dimension of the received signal vector;

[0040] - Use the data stream selection information included in the signaling message to select multiple data streams corresponding to the components of the transformed vector.

[0041] - Apply a quantization operation to the selected data stream, the quantization operation providing a quantized vector.

[0042] The method includes transmitting the quantized vector from the radio unit to the distributed unit.

[0043] In this context, the sum of the number of data streams selected at each O-RU on all radio units of the communication system 100 is greater than or equal to the number of user equipment K and the number of antennas of the user equipment. Attached Figure Description

[0044] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate various embodiments of the invention and, together with the general description of the invention given above and the detailed description of the embodiments given below, serve to explain the embodiments of the invention.

[0045] - Figure 1 A communication system using an O-RAN architecture is described according to some embodiments of the present invention;

[0046] - Figure 2 This is a diagram illustrating the structure of an O-RU according to some embodiments;

[0047] - Figure 3 This is a diagram illustrating the structure of an O-DU according to some embodiments;

[0048] - Figure 4This is a flowchart describing the method of sending a data stream from the O-RU to the O-DU;

[0049] - Figure 5 This is a flowchart describing the method by which the O-DU processes the data streams received from all O-RUs;

[0050] - Figure 6 It is a graph describing the symbol error rate (SER) as a function of signal-to-noise ratio (SNR) obtained using different embodiments of the present invention compared to conventional schemes;

[0051] - Figure 7 This is a graph describing the symbol error rate (SER) as a function of the total number of quantization bits per O-RU obtained using the CNP implementation of the present invention, compared to conventional schemes. Detailed Implementation

[0052] Figure 1 This refers to a wireless communication system 100 implemented according to the O-RAN architecture.

[0053] O-RAN is configured to provide network services to one or more user equipment (UE)3 via radio signals.

[0054] Each user device 3 can be a stationary or fixed wireless portable device configured to communicate using a wireless connection, such as a cellular phone, handheld device, tablet computer, laptop computer, or smartphone.

[0055] The O-RAN architecture includes one or more multi-radio units: O-RAN radio unit O-RU 5 (also known as radio head), O-RAN distributed unit O-DU 6 (also known as "central processing unit" or CP), and O-RAN centralized unit O-CU 7.

[0056] The O-RU 5 can provide wireless services to one or more user equipment 3.

[0057] O-RU 5 may include one or more antennas. Each user equipment 3 may have at least one antenna.

[0058] In the considered uplink cloud RAN model, it is assumed that there are N t K remote users with 3 transmitting antennas independently transmit their messages to N remote O-RUs 5 using their respective user equipment 3. Each O-RU 5 is equipped with N r antennas (N) r >>N t ).

[0059] Each O-RAN radio unit (O-RU) 5 is connected to the O-RAN distributed unit (O-DU) 6 via an O-RAN fronthaul link (FH) 4. Fronthaul link 4 is a noise-free fronthaul link, assumed to have a finite capacity C. i This means that signals received at O-RU 5 need to be compressed in a distributed manner to meet capacity constraints before being forwarded to O-DU 6.

[0060] The communication system 100 includes a coordination system 20 that coordinates the exchange between O-RU 5 and O-DU 6. By further reducing the dimensionality of the data stream sent from O-RU to O-DU 6, the coordination system 20 implements a coordinated compression scheme between O-RU 6, which enhances the decoding performance at O-DU 6.

[0061] Communication system 100 can be used in many applications, such as MU-MIMO communication, wireless sensor networks, or NR5G systems.

[0062] In one implementation, the communication network 100 may implement MIMO-based techniques (such as MIMO or MU-MIMO) for transmitting data packets (also referred to as data “frames”) in the O-RAN network.

[0063] The coordination system 20 is configured to perform compression of signals received by the O-RU 5 in a coordinated manner by using signaling messages.

[0064] The i-th forward link 4 between the i-th O-RU and O-DU is considered to have a separate and identical capacity constraint C. i It is less than or equal to the capacity threshold C(C i ≤C). A message sent by user i equipped with user equipment 3 is represented as x. i And it belongs to a finite constellation.

[0065] Signal x sent by user equipment i i It is received by each radio unit O-RU j 5.

[0066] The signal y received by the first O-RU j 5 j It can be represented as:

[0067]

[0068] In equation (1), Let represent the channel matrix between O-RU j and user i, and This represents the received noise at point j in the O-RU.

[0069] Equation (1) can also be written as:

[0070] y j =H j x+z j

[0071] H represents i Representing the channel matrix And H ij This represents the channel vector between user i and O-RU j.

[0072] In the following description of some embodiments of the present invention, the communication system 100 is considered to include K user equipments 3 and N O-RUs 5, each user equipment 3 including N t One antenna, O-RU 5 has N r N receiving antennas r >>K).

[0073] Figure 2 This indicates the structure of O-RU 5 according to some embodiments of the present invention.

[0074] Each O-RU n 5 may include a transformation unit 50, which is configured to process the received signal y n Apply transformation operations To receive the signal vector y n The signal is transformed into a vector, thereby reducing the dimensionality of the received signal.

[0075] In one implementation, it is applied to the received signal vector y n Transformation operations Including through the channel matrix H n Hermitian transpose H of the channel matrix from all users to O-RU n n H To scale the received signal vector y n It provides the transformed vector Superscript (.) H This represents the Hermitian transpose operator.

[0076] Transformed vector The dimension is KN t The received signal vector y n The dimension is N r ,in, Therefore, the conversion unit 50 will receive the signal y n Transformed into a vector with reduced dimension.

[0077] To facilitate understanding of the present invention, the following description of some embodiments will refer to the received signal vector y. nTransform into the transformed vector The transformation operation is performed on the received signal vector y, but those skilled in the art will readily understand that other transformation operations can be applied to it. j This provides a transformed vector with reduced dimensionality that carries information about the received signal vector.

[0078] Each O-RU n may also include a data stream selection unit 52, which is configured to select a transformed vector using information included in a signaling message received from the coordination system 20 in response to a signaling message received from the coordination system 20. K n Each component.

[0079] In one implementation, the coordination system 20 can be configured to send N numbers (K1,...K5) to N O-RUs 5. n ,…,K N The signaling message contains various numbers K. n This represents the number of data streams that will be sent from the nth O-RU 5 (O-RUn) to the O-DU 6 (each data stream corresponds to a transformed vector). The amount ), making

[0080] In response to receiving a signaling message, O-RU n can extract the value K from the signaling message. n , value K n This indicates the number of components to be sent from O-RU n to O-DU 6. Then, the data stream selection unit 52 of O-RU n can use the number K. n To determine K to O-DU 6 n Therefore, O-DU 6 will receive data from all O-RU 5 (O-Ru 1,...O-Ru n,...O-Ru N). There are 1 data stream, of which:

[0081]

[0082] In one implementation, the coordination system 20 can be configured to measure eigenvalues ​​of the covariance channel matrix at each O-RU 5 or O-DU 6. These eigenvalues ​​can be exchanged between O-RU 6 and used by the data stream selection unit 52 of each O-RU 5 to determine the data stream to be sent from each O-RU 5 to the O-DU 6.

[0083] In some implementations, the coordination system 20 can be configured to use a threshold TH for exchanging eigenvalues ​​among O-RUs 5 using signaling messages. The data stream selection unit 52 of each O-RU 5 also uses this threshold to determine which data streams should be sent from each O-RU to O-DU 6. The data stream selection unit 52 of the O-RU n receiving the signaling messages can be configured to select the components of the transformed vector corresponding to the eigenvalues ​​of the covariance channel matrix that are above the threshold TH.

[0084] In some implementations, the signaling messages sent to each O-RU n may include a length of N. r Selection vector V n , where N r This indicates the number of receiving antennas at the O-RU. The matrix is ​​diag(V n ) indicates that vector V is included on its main diagonal. n A diagonal matrix of elements.

[0085] In one implementation, if the i-th component of the transformed vector must be sent from O-RU n to O-DU 6, then vector V n The i-th component V n (i) has a value equal to 1. Conversely, if the i-th component of the transformed vector will not be sent from O-RU n to O-DU 6, then vector V n The i-th component V n (i) has a value equal to 0.

[0086] As used in this paper, the i-th component of a vector refers to the element of the vector defined by index i. Similarly, the components of a matrix refer to the matrix elements defined by their row and column indices.

[0087] In some implementations, it can be based on the covariance channel matrix H n H H n The eigenvalues ​​are used to determine the selection vector V n .

[0088] In one implementation, if the i-th eigenvalue of the covariance channel matrix is ​​greater than or equal to the threshold TH, then the vector V n The i-th coefficient V n (i) has a value equal to 1; conversely, if the i-th eigenvalue of the covariance channel matrix is ​​below the threshold TH, then the vector V n The i-th coefficient V n (i) has a value equal to 0.

[0089] Advantageously, signaling messages are used to send to different O-RUs 5 (so that O-RU n receives vector V). nSend selection vector (V1,...V) n ,...V N This ensures that the nth O-RU 5 will only send K. n A data stream, numbers (K1,...K) n ,…,K N ) makes Where K n This indicates the number of data streams that will be sent from O-RU n to O-DU 6.

[0090] To ensure that different O-RU 5s do not send the same data stream to O-DU 6, the vector (V1,...V n ,...V N ) is determined to be such that:

[0091]

[0092] The data stream selection unit 52 of the O-RU n can execute the matrix diag(V) according to equation (4.1). n ) and the transformed vector H n H y n Matrix multiplication between them is used to select the K to be sent to O-DU 6. n One data stream:

[0093] r n =diag(V n )H n H y n (4.1)

[0094] In equation (4), r n A vector representing the selected data stream.

[0095] Then, O-RU n will select the vector r of the data stream. n Applied to the quantizer to apply the quantization operation Q(.), which provides the quantized vector.

[0096]

[0097] Then, the quantized vector is sent from O-RU n to O-DU 6. Therefore, the quantized vector includes K of the transformed vector. n K corresponding to each component n A data stream.

[0098] Figure 3 This indicates the structure of O-DU 6 according to some implementation methods.

[0099] like Figure 3 As shown, O-DU 6 may include a reconstruction unit 60, which is configured to reconstruct the vector received from all O-RU 5. To reconstruct the received signal y. In one implementation, the reconstruction unit 60 of O-DU 6 can be configured to reconstruct the received signal y based on the quantized vector received from all O-RU 5 (O-RU 1,...,O-RU n,...O-RU N). To perform the reconstruction of the received signal vector, which provides the reconstructed vector. For example, the reconstruction operation can be performed on the quantized vector according to equation (4.3). Summation:

[0100]

[0101] The O-DU 6 may also include a decoder 62 configured to decode the reconstructed vector r to provide an estimate of the received signal. Decoder 62 can be, for example, an MMSE decoder or a ZFDE decoder.

[0102] By sending a reduced number of data streams from each O-RU 5 to the O-DU and concatenating all data streams received from different O-RU 5s in the O-DU 6, the coordination system 20 reduces quantization noise compared to conventional schemes. In fact, this is achieved by allowing vector selection at O-RU n. The best N in j Entries (N) j <KN t Instead of sending the entire vector. The coordination system 20 enables the mitigation of degradation caused by quantization performed by quantizer 54. This also avoids receiving N times more observations at O-DU 6 than the number of user messages to be decoded (i.e., the dimension of x).

[0103] In one implementation, coordination system 20 may be a centralized coordination system. Alternatively, coordination system 20 may be a distributed coordination system.

[0104] In a centralized implementation, signaling messages can be constructed by O-DU 6 and sent to each O-RU 5 (O-Ru1,...,O-Ru n,...O-Ru N).

[0105] In this centralized implementation, assuming that the channel state information (CSI) is fully known at O-DU 6, the signaling message can be related to the CSI.

[0106] In a centralized implementation, O-DU 6 can determine the selection vector (V1,...V) associated with the corresponding O-RU (O-Ru 1,...,O-Run,...O-Ru N) based on information related to the covariance channel matrix (e.g., eigenvalues ​​or determinants of the covariance channel matrix). n ,...V N ).

[0107] O-DU 6 can send signaling messages to each O-RU 5 one by one, or alternatively send signaling messages to all O-RUs simultaneously or in a specific or arbitrary order.

[0108] Signaling messages may rely on Channel State Information (CSI), which includes information related to the channel matrix H. In some implementations, signaling messages can be sent to O-RU 5 dynamically. In such implementations, O-DU 6 can dynamically determine the selection vector (V1,...V1) independent of the signals received by the O-RU from the user. n ,...V N And send to different O-RUs (O-Ru 1,...,O-Ru n,...O-Ru N). Specifically, because {V n The determination of the vector depends on CSI, so {V} n The transmission of the vector can be accomplished in a manner characterized by constant variations in the CSI. Therefore, when the O-RU 5 receives a signal from the user, the O-RU 5 already knows what it is used to select K. n The selection vector V that must be applied for a data stream n This allows it to perform data stream selection and quantization of the selected data stream without waiting to receive signaling messages from O-DU 6.

[0109] Therefore, the transformed signal is not sent from each O-Ru n to O-DU 6. All KN t Instead of individual components, the coordination system 20 uses a diagonal matrix diag(V) j To choose K in n Each component (K) n The coordination system 20 uses a subset of selected data streams to send data from each O-Run 5 to O-DU 6 in a coordinated manner. Therefore, the coordination system 20 can better utilize quantization resources and improve the reconstruction quality of O-DU 6.

[0110] In some implementations, the coordination system 20 may also be configured to determine the vector V to be sent to O-Ru n. n This makes the vector of the selected data stream... It is sparse, which allows for better exploration of fronthaul link resources. Utilizing vectors... This sparse structure does not require vectors Quantization is performed on zero entries, and high accuracy can be expected for non-zero entries after quantization. Sparse vectors are used. This can mitigate the degradation caused by quantization noise.

[0111] To maintain decoding performance, it is assumed that the number of observations received at O-DU 6 from all O-RU 5 is greater than or equal to the number of user messages to be decoded.

[0112] Coordination system 20 can be configured to define a selection vector {V} n} satisfies the following two constraints:

[0113] - The first constraint includes maintaining H T y n The kth entry can be eliminated using zero.

[0114] - The second constraint includes ensuring the matrix The matrix is ​​full rank, and therefore the matrix is ​​full rank. It is reversible.

[0115] In the following description of some embodiments of the invention, an uplink O-RAN architecture with Gaussian input at the user and Gaussian quantization at the O-RU5 is considered. Furthermore, an exemplary case of a communication system 100 comprising two O-RUs (N=2) is considered to aid in understanding the invention, although those skilled in the art will readily understand that the invention is applicable to communication systems 100 comprising any number of N O-RUs 5.

[0116] Since both the signal and noise follow a Gaussian distribution, when O-RU j sends the quantized signal to O-DU 6... Then, the sum rate optimization problem, which characterizes the achievable rate region, can be expressed as (5) and (6):

[0117]

[0118]

[0119] In expressions (5) and (6), This represents the covariance matrix of a user message with a transmission power P. Let denot Λ represent the quantization variance at point O-RU j 5, and Λ q Defined as:

[0120]

[0121] Since the global optimal solution of (5) is difficult to obtain, the following practical assumptions are made to evaluate the achievable rate:

[0122] -When H j When each entry is iid (an abbreviation for "independent and identically distributed") and the number of receiving antennas is sufficiently large (N) r >>1), According to the law of large numbers, it can be checked Each entry follows the same distribution. Therefore, it is assumed that for Each entry introduces the same level of quantization, that is,

[0123] - To simplify the determinant operator, consider the low-resolution region, i.e., the fronthaul link capacity tends to zero (C→0).

[0124] By selecting vector V at each O-RU j 5 j Determine the vector of the selected data stream And in sending the vector r of the selected data stream to O-DU 6 j Previously, it was quantified, and the coordination system 20 was able to achieve a higher performance than O-RUj. The traditional approach of quantizing and forwarding to O-DU 6 has good speed and efficiency.

[0125] In order to coordinate the performance of the compression system 20 with its O-RU j pair Quantitative comparisons of the performance of traditional systems can be used to compare the performance of... The quantization achievable and rate of quantization The quantization achievable sum rate.

[0126] To determine the sending Considering the small values ​​of ∈, the determinant operator can be approximated according to equation (7):

[0127]

[0128] Therefore, in the low-resolution region with small C, the left-hand side of the constraint defined by (6) can be rewritten as equation (8):

[0129]

[0130] To maximize the sum rate defined by equation (5), the inequality constraints can be relaxed to equality constraints, and thus the quantization level is determined. It can be approximated by equation (9):

[0131]

[0132] By integrating equation (9) into equation (5), the sum rate can be simplified and rewritten as (10):

[0133]

[0134] When sending the vector of the selected data stream Instead At that time, and by restating the optimization problem defined by expressions (5) and (6) with the additional term Vj, where Vj is the j-th choice vector including 0 or 1 components, the achievable rate (10) can be derived in a similar manner and rewritten as:

[0135]

[0136] Due to the matrix daig(V) j ) is a vector with 0 or 1 components, so for a vector with diagonal terms x kk Any square matrix X, where k∈{1,…,KN} t}, V j XV j H The trace can be considered as {x kk The sum of subsets of}, i.e.

[0137] Therefore, for any matrices X and Y, inequality (12) is verified:

[0138]

[0139] Therefore, this shows that by appropriately choosing the Vj(k) components, there exists Multiple combinations enable transmission in the asymptotic region with small fronthaul link capacity. The achievable rate can be better than the transmission rate. The rate, in the example N=2.

[0140] Therefore, for the rates achievable in low-resolution regions, it is advantageous to send fewer data streams per O-RU.

[0141] In some implementations, the coordination system 20 may include a selection vector determination unit 63, which is configured to appropriately select a selection vector {V}. n This is to enhance the decoding performance at O-DU 6.

[0142] In a centralized implementation of the coordination system 20, the selection vector determination unit 63 can be included in the O-DU 6, such as... Figure 4 As shown.

[0143] To facilitate understanding of some embodiments of the present invention, the selection vector {V} will be described with reference to an exemplary implementation of a communication system 100 including two O-RUs 5. j The determination of} is not clear, but those skilled in the art will readily understand that the present invention is more generally applicable to any number of O-RUs.

[0144] Depending on the information required at the O-RU, the selection vector determination unit 63 can determine the selection vector {V} according to two determination schemes. j}:

[0145] -In the first determined scheme, a set of selection vectors {V} j It depends only on the statistical information of the channel matrix H.

[0146] - In the second determined scheme, a set of selection vectors {V j It depends only on the instantaneous realization of the channel matrix H.

[0147] According to the first determined scheme, the group selects vector {V} j} can be fixed and is independent of instantaneous channel information.

[0148] In this scenario, although there is no need to exchange channel information, the number K of data streams to be sent from each O-RU j 5 to O-DU 6 can be predefined. j And the set of selection vectors V = {V} can be pre-defined. j Non-zero entries in}. In one implementation, the observation with the strongest average power across all channel implementations can be selected. For example, if If the k-th and m-th entries have the strongest average power across all channel implementations, then each selection vector Vj can be set to be defined such that the components Vj(k) = 1 and Vj(m) = 1 can be maintained for the two strongest entries.

[0149] It should be noted that when the entries of H are iid (independent and identically distributed), Each entry has the same power. In this case, the selection of the element equal to 1 is randomized. k can be randomly chosen such that V... j (k) = 1 (because all entries have the same power).

[0150] According to the second determined scheme, a set of selection vectors V = {V j It depends on instantaneous channel information. The non-zero entries correspond to the data streams (observations) to be quantized and forwarded to O-DU 6. In one implementation of the second determination scheme, the group selection vector V = {V} can be determined based on the channel matrix H. j}

[0151] Specifically, the selection vector V = {V} can be determined based on RSNR priority or RSNRP (RSNRP is an abbreviation for "Receive Signal-to-Noise Ratio Priority"). j In such an implementation, consider the marking. The second determination scheme implemented by the vector determination unit 63 may include KN at the first O-RU 5. t The optimal K1 channel vectors are selected from the channel vectors to maximize the received signal-to-noise ratio (SNR), as expressed in (13):

[0152]

[0153] By defining the indices of the K1 channel vectors with the best received SNR as if Then the vector determination unit 63 can determine the elements of the first selection vector V1 as V1(k) = 1, and if Then V1(k′) = 0. The second selection vector V2 can be determined such that:

[0154]

[0155] Therefore, the components of the second selection vector V2 make when When V2(k) = 1, and when When V2(k′) = 0. According to the RSNRP scheme, priority index Exchanges can be made between O-RU 5s, allowing two O-RU 5s to send selected data streams in a coordinated manner. In one implementation, auxiliary signaling messages between O-RU 5s can be used to perform priority indexing between O-RU 5s. The exchange.

[0156] More generally, if the communication system 100 includes N O-RUs, then the optimal K that maximizes the received SNR (signal-to-noise ratio) can be selected. n A data stream is used to determine the vector {V} n}, as expressed in (13). Furthermore, the following constraint is satisfied: for all k-th data streams, V n (k) The sum of n = 1 to N is greater than or equal to 1.

[0157] Alternatively, according to the second determination scheme, the selection vector V = {V} can be determined based on the condition number priority (CNP). j}

[0158] Assuming O-DU 6 has perfect knowledge of H, then vector determination unit 63 can be configured to select a combination of V1 and V2 in a manner that maximizes the ratio between the following:

[0159] -matrix The determinant, and

[0160] -matrix The condition number is defined by the following formula:

[0161]

[0162] λ min Let λ represent the smallest eigenvalue, and λ be the smallest eigenvalue. max This represents the largest eigenvalue.

[0163] Depend on Defined matrix The condition number CNP is given by expression (14):

[0164]

[0165] Significant gains can be achieved by selecting the data streams to be sent by each O-RU, ensuring a well-tuned matrix at O-DU 6. Considering the degradation caused by quantization noise, each O-RU will send fewer data streams, especially in low-resolution regions. Therefore, the total number of data streams sent from all O-RUs 5 to O-DU 6 can be set equal to the number of user messages to be decoded, such that selection vectors V1 and V2 satisfy:

[0166]

[0167] Therefore, CNP can be rewritten according to expression (15):

[0168]

[0169] In this CNP-based implementation, O-DU 6, possessing perfect knowledge of all channel information, can determine the selection vector {V}. n}, and uses signaling messages to send each selection vector V to the corresponding O-RU n. n For example, with N=2, and all channels known, O-DU 6 can jointly determine vectors V1 and V2, and send V1 to the first O-RU 5 and V2 to the second O-RU 5.

[0170] More generally, for a communication 100 comprising any number N O-RUs, this can be achieved by making Maximizing the ratio between the determinant and its condition number determines the choice vector V. n(For 1≤n≤N).

[0171] Depending on the vector determination scheme used by vector determination unit 63, this implementation can involve different features. For example, a first determination scheme can be implemented offline by vector determination unit 63 without additional signaling, because in this implementation, vector V j It relies solely on channel statistics. Conversely, the second determination scheme can be implemented online by the vector determination unit 63 by exchanging instantaneous channel information. Furthermore, the second determination scheme involving RSNRP is likely less difficult to implement than the one involving CNP, because the information to be exchanged for this scheme can be exchanged in a distributed manner among the O-RUs 5, which is different from the need for a centralized entity (O-DU) to jointly determine V. j The CNP schemes sent to each O-RU 5 are different.

[0172] The performance of the coordination system 20 according to an embodiment of the present invention has been compared with that of the system for... A comparison was made with the traditional approach of quantizing and forwarding to O-DU6.

[0173] To estimate decoding performance, consider an example of a communication system 100 consisting of four single-antenna users and two O-RU5s equipped with ten receive antennas, comparing the symbol error rate (SER) of different technologies. In the considered example, element-wise uniform quantization is used at each O-RU5. The fronthaul link capacity can be represented by R, where R represents the total quantization bits per O-RU5. In the considered example, it is assumed that the total quantization bits of each O-RU are uniformly allocated to all non-zero entries. It is assumed that user messages follow a 16QAM constellation. It is assumed that all channels are independent and identically distributed (iid) and follow a normal distribution.

[0174] Furthermore, in the example considered, the decoder 62 used at O-DU 6 is a ZF decoder, which is configured to multiply the (pseudo)inverse matrix to reconstruct the user message. The signal received from O-RU j at O-DU 6 is used... Let i ∈ {Con,Nov}, where Con represents sending to O-DU. The conventional scheme for all components, and Nov indicates that, according to an embodiment of the invention, it is configured to send to the O-DU. The coordination system 20.

[0175] The ZF decoder 60 can be represented by equation (16) for the conventional system Con, and in the embodiments of the present invention, it can be represented by equation (17):

[0176]

[0177]

[0178] Therefore, embodiments of the present invention provide a coordination system 20 that can perform distributed compression at the O-DU, enabling the O-DU to better jointly decode user messages.

[0179] The coordination system 20 enables the transmission of as much useful information as possible from the O-RU to the O-DU 6.

[0180] Advantageously, the compression system 100 uses signaling messages to transmit information sent or exchanged between O-RUs on different channels to take advantage of the correlation between signals received at different O-RUs. These signaling messages form a coordination key that enables all O-RUs to compress the received signals cooperatively, thereby enhancing the performance of the coordinated compression scheme among multiple O-RUs and improving the joint decoding performance at the Open Distributed Unit (O-DU) 6.

[0181] The coordination system 20 enables the determination of which data streams are sent by each O-RU and how to combine / cascade all these data streams received at O-DU6 to better decode user messages.

[0182] This prevents the large degradation of message reconstruction at O-DU 6 that may occur in the prior art due to quantization, especially in the case of large-scale MIMO.

[0183] The embodiments of the present invention enable the signals received at each O-RU to be transformed into smaller vectors and transmitted to O-DU 6 in a coordinated manner to reduce noise caused by quantization while ensuring decoding performance at O-DU 6.

[0184] The coordination system 20 achieves coordination between O-RUs through information exchange between O-RUs (distributed scheme) or by using signaling messages to feed back information from O-DU6 to O-RUs (centralized scheme). Using signaling messages, each O-RU can collaboratively determine how much data stream to send to O-DU6 and which data streams to send to O-DU6, enabling O-DU to achieve better decoding performance in the presence of quantization noise.

[0185] Advantageously, the structure of the signaling message containing data stream selection information is simple, resulting in low overhead.

[0186] Figure 4 This is a flowchart illustrating the coordination method performed at O-RU j for sending the selected data stream to O-DU 6.

[0187] In step 400, user messages are received by O-RU 5, and the messages received at each radio unit 5 are represented by signal vectors.

[0188] In step 402, a signaling message including data stream selection information is received at O-RU j.

[0189] In step 404, a transformation operation is applied to the signal vector received by O-RU j 5, which provides a transformed vector. This transformation is defined as reducing the dimension of the received signal vector.

[0190] In step 406, data stream selection information (e.g., selection vector V) included in the signaling message is used. n This allows you to select multiple data streams that correspond to the components of the transformed vector.

[0191] In step 408, a quantization operation is applied to the selected data stream, which provides a quantized vector.

[0192] In step 410, the quantized vector is sent from O-RU j to distributed unit 6.

[0193] Figure 5 This is a flowchart describing the decoding process of the received signal vector performed at O-DU 6.

[0194] In step 500, O-DU 6 receives the quantized vector from all O-RU 6. Thus, O-DU 6 receives K from each O-RU n. n A data stream.

[0195] In step 502, the summation of all quantized vectors received at O-DU 6 is performed to reconstruct the received signal.

[0196] In step 504, the reconstructed signal is decoded to determine an estimate of the signal sent by the user.

[0197] In a centralized scheme, the coordination method may also include determining, at O-DU 6, data flow selection information to be sent to all O-RUs (e.g., selection vector {V}). n}), and send a signaling message containing data stream selection information.

[0198] Figure 6 This is a graph depicting the symbol error rate (SER) as a function of the signal-to-noise ratio (SNR) compared to conventional schemes using different embodiments of the present invention. SNR is determined by... Definition. In Figure 6 In the picture:

[0199] -curve This represents the SER obtained using an exemplary implementation, in which a first deterministic scheme is used to determine the selection vector {V}. n};

[0200] -curve This represents the SER obtained using an exemplary implementation, in which the RSNRP determination scheme is used to determine the selection vector {V}. n};

[0201] -curve This represents the SER obtained using an exemplary implementation, in which the CNP determination scheme is used to determine the selection vector {V}. n};

[0202] -curve This represents the SER obtained using a conventional approach, which involves sending all components of the transformed vector from each O-RU to O-DU6.

[0203] like Figure 6 As shown, by the curve The symbol used for sending The conventional approach exhibits smaller reconstruction errors at relatively low SNRs because transmitting more information can be beneficial when the dominant noise is additive noise received at O-RU 5 rather than quantization noise. More importantly, when the SNR is above 3 dB, all vector determination schemes are observed to outperform the conventional approach.

[0204] Figure 6 It also shows that when more information is needed at O-RU 5, the curve... and The representation used to determine the selection vector V n Different technologies offer lower SER.

[0205] Therefore, when the channel noise received at O-RU 5 is not dominant compared to the quantization noise, the embodiments of the present invention are more effective in reconstructing user messages.

[0206] Figure 7 This is a graph describing the symbol error rate (SER) of the CNP implementation according to the present invention as a function of the total number of quantization bits per O-RU, compared to conventional schemes.

[0207] Figure 7 The impact of fronthaul link capacity is depicted, and a comparison with different quantization numbers of SER is provided.

[0208] exist Figure 7 In the picture:

[0209] -curve This represents the SER obtained using an exemplary implementation, in which the CNP determination scheme is used to determine the selection vector {V}. n};

[0210] -curve This represents the SER obtained using a conventional scheme, which involves sending all components of the transformed vector from each O-RU to O-DU 6.

[0211] exist Figure 7 In this context, the gain from the vector determination scheme is expressed as a function of the total number of quantized bits per O-RU 5 (i.e., the fronthaul capacity). In low and medium resolution regions, the CNP determination scheme can reduce the SER by up to one-tenth, since quantization noise substantially affects decoding performance. However, when the number of quantized bits is sufficiently large (e.g., 32 bits), quantization noise becomes negligible. Although conventional methods can achieve lower SER, the coordination system 20 can provide significant improvements when quantization noise is not negligible.

[0212] The coordination system 20 according to embodiments of the present invention thereby enhances the SER performance in cloud RAN systems. It improves efficiency compared to conventional schemes even in asymptotic states assuming very low throughput. Furthermore, significant gains can be obtained by using selection vectors in low and medium resolution regions.

[0213] It should be noted that although some embodiments of the present invention have been described with reference to a centralized coordination system 20, in which signaling messages and data stream selection information are determined and sent from O-DU 6, the present invention is also applicable to a distributed coordination system 20.

[0214] In this distributed implementation of the invention, instead of receiving signaling messages from O-DU 6, each O-RU 5 can exchange eigenvalues ​​of the covariance channel matrix with other O-RU 5, or more generally, exchange data stream selection information.

[0215] The embodiments of the present invention may take the form of including only software, only hardware, or both hardware and software elements.

[0216] For example, the methods described herein, the O-RU 5 and / or O-DU 6 according to embodiments of the present invention can be implemented in hardware and / or software.

[0217] Furthermore, the methods described herein can be implemented by computer program instructions provided to a processor of any type of computer to produce a machine having a processor that executes instructions to perform the functions / actions specified herein. These computer program instructions can also be stored in a computer-readable medium that can instruct a computer to operate in a particular manner. For this purpose, computer program instructions can be loaded onto a computer to cause a series of operational steps to be performed, thereby producing a computer-implemented process, such that the executed instructions provide a process for implementing the functions specified herein.

[0218] It should be noted that, according to embodiments of the present invention, the functions, actions, and / or operations specified in flowcharts, sequence diagrams, and / or block diagrams can be reordered (e.g., Figure 4 Step 402 may be performed before step 400 or after header 404, and may involve serial and / or concurrent processing. Furthermore, any flowchart, sequence diagram, and / or block diagram may include more or fewer blocks than those shown in embodiments of the invention.

[0219] While embodiments of the invention have been described by way of various examples, and while these embodiments have been described in considerable detail, the applicant does not intend to limit the scope of the appended claims or in any way to such details. Other advantages and modifications will be apparent to those skilled in the art. Therefore, the invention, in its broader aspects, is not limited to the specific details, representative methods, and illustrative examples shown and described.

Claims

1. A communication system (100) having an open radio access network architecture, the communication system comprising K user equipments (3) configured to exchange messages through the communication system, each user equipment comprising a plurality of transmit antennas, the communication system comprising one or more radio units (5) and a distributed unit (6), each radio unit (5) being connected to the distributed unit via a fronthaul link (4) having a given capacity, wherein, The communication system includes a coordination system (20) configured to coordinate the exchange between the radio units (5) and the distributed unit (6) in response to a message received by the radio unit (5) from the user equipment (3), wherein the message received at each radio unit (5) is represented by a signal vector, and wherein the coordination system (20) is configured to send a signaling message including data stream selection information to each radio unit (5), wherein each radio unit includes: A transformation unit (50) is configured to apply a transformation operation to a signal vector received by the radio unit (5), the transformation operation providing a transformed vector, the transformation operation being defined as reducing the dimension of the received signal vector; A data stream selection unit (52) is configured to use the data stream selection information included in the signaling message to select multiple data streams corresponding to the components of the transformed vector. A quantizer, configured to apply a quantization operation to a selected data stream, the quantization operation providing a quantized vector. The radio unit (5) is configured to transmit the quantized vector to the distributed unit (6). Among them, on all radio units connected to the distributed unit (6), the sum of the number of data streams selected at each radio unit (5) is greater than or equal to the product of the number of user equipment K and the number of antennas of the user equipment.

2. The communication system according to claim 1, wherein, The distributed unit (6) includes a reconstruction unit (60) and a decoder (62). The reconstruction unit (60) is configured to perform a reconstruction of the received signal vector based on the quantized vector received from all the radio units, the reconstruction providing a reconstructed signal vector. The decoder (62) is configured to decode the reconstructed signal vector to determine an estimate of the signal transmitted by the user equipment (3).

3. The communication system according to claim 1, wherein, The data stream selection information received by a given radio unit via signaling messages includes a selection vector V associated with the given radio unit O-RU n. n The selection vector V n It has a length N corresponding to the number of receiving antennas included in the associated radio unit. r The data stream selection unit (52) of a given radio unit O-RU n is configured to extract the selection vector V associated with the given radio unit O-RU n from the signaling message. n Furthermore, matrix multiplication is performed between the diagonal matrix and the transformed vector, the diagonal matrix comprising the selected vector V. n The components form a diagonal, the matrix multiplication provides a vector of the selected data stream, and the quantizer of the given radio unit is applied to the vector of the selected data stream.

4. The communication system according to claim 3, wherein, The selection vector associated with each radio unit is determined based on channel state information.

5. The communication system according to claim 4, wherein, If the i-th component of the transformed vector must be transmitted from the given radio unit to the distributed unit (6), then the selection vector associated with the given radio unit... The i-th coefficient V n (i) has a value equal to 1, and conversely, has a value equal to 0 if the i-th component of the transformed vector will not be transmitted from the given radio unit to the distributed unit (6).

6. The communication system according to claim 3, wherein, The one or more radio units include N radio units, and the data stream selection information includes a set of N selection vectors. Each selection vector is associated with a corresponding radio unit, wherein each selection vector V n It is determined based on the eigenvalues ​​of the covariance channel matrix.

7. The communication system according to claim 3, wherein, A set of selection vectors associated with all radio units (5) Determined to make: ， Wherein, N represents the number of radio units.

8. The communication system according to claim 3, wherein, The selection vector is determined by the distributed unit.

9. The communication system according to claim 3, wherein, The selection vector associated with the radio unit depends only on the statistical information of the channel matrix.

10. The communication system according to claim 9, wherein, If the k-th and m-th entries of the covariance channel matrix of the transformed vector have the strongest average power across all channel implementations, then each selection vector V with a given index j j Defined such that the selection vector V j The component Vj(k) is equal to the value 1 and the selection vector V j The component Vj(m) is equal to the value 1.

11. The communication system according to claim 3, wherein, The selection vector associated with the radio unit depends only on the instantaneous implementation of the channel matrix.

12. The communication system according to claim 11, wherein, The selection vector associated with the radio unit is determined based on the received signal-to-noise ratio priority.

13. The communication system according to claim 12, wherein, The coordination system (20) is configured to send auxiliary signaling messages between the radio units to exchange priority indices.

14. The communication system according to claim 10, wherein, The selection vector associated with the radio unit is determined based on the condition number priority of the channel matrix.

15. A coordination method implemented in a communication system having an open radio access network architecture, the communication system comprising K user equipments (3) exchanging messages through the communication system, each user equipment comprising multiple antennas, the communication system comprising one or more radio units (5) and a distributed unit (6), each radio unit (5) being connected to the distributed unit via a fronthaul link (4) having a given capacity, wherein, The coordination method includes coordinating the exchange between the radio unit (5) and the distributed unit (6) of the system in response to the radio unit (5) receiving a message sent from the user equipment (3), wherein the message received at each radio unit (5) is represented by a signal vector, wherein the method includes sending a signaling message including data stream selection information to each radio unit (5), wherein the method further includes at each radio unit: A transformation operation is applied to the signal vector received by the radio unit (5), the transformation operation providing a transformed vector, the transformation being defined as reducing the dimension of the received signal vector; Multiple data streams corresponding to the components of the transformed vector are selected using the data stream selection information included in the signaling message. A quantization operation is applied to the selected data stream, and the quantization operation provides a quantized vector. The method includes transmitting the quantized vector from the radio unit (5) to the distributed unit (6). Among them, on all radio units connected to the distributed unit (6), the sum of the number of data streams selected at each radio unit is greater than or equal to the product of the number of user equipment K and the number of antennas of the user equipment.

Images