Downlink robust distributed precoding implementation system and implementation method of cellular-free wireless access network

Abstract

The invention discloses a downlink robust distributed precoding implementation system and implementation method of a cellular-free wireless access network. The system comprises a first unit, a second unit, an Ethernet switching unit and a third unit of each frequency band, the first unit completes sending and receiving of wireless radio frequency signals of all frequency bands, digital-to-analog and analog-to-digital conversion and waveform signal processing of a physical channel. The second unit completes channel parameter estimation, downlink precoding calculation and precoding; the third unit calculates a downlink statistical precoding according to the statistical information sent by the second unit; and the Ethernet switching unit is used for connecting different units and carrying out information interaction. According to the method, the problems of high-performance implementation of existing non-cellular large-scale MIMO distributed precoding and robustness to phase noise can be solved.

Description

Technical Field

[0001] The present invention relates to the field of wireless communication technology, and in particular to a downlink robust distributed precoding implementation system and implementation method for a non-cellular wireless access network. Background Art

[0002] Cellular networks include macro cells with wide-area coverage and micro cells and pico cells with local coverage. Dense network layout can significantly improve network throughput, but the edge user problem has never been effectively solved. Early distributed antenna systems can and cannot solve this problem well. Cell-free networks, as a distributed implementation version of massive antennas (MIMO), make it possible to solve the interference problem of cell edge users. In cell-free wireless access networks, distributed signal processing can realize cell-free massive MIMO cooperative transmission.

[0003] However, in terms of implementation, first of all, there is a general trade-off between the distribution capability and spectrum efficiency of non-cellular networks: fully centralized processing can effectively suppress interference and noise to achieve a throughput that is difficult to achieve in traditional cellular networks, but it places extremely high computing power requirements on the central processing unit (third unit) and requires a large amount of backhaul costs in each time-frequency resource block; distributed downlink processing can disperse the calculation to distributed baseband units or even wireless radio frequency units, greatly reducing the cost of backhaul, but the current distributed downlink processing performance is difficult to compare with the fully centralized approach. Secondly, for downlink scenarios, the current common practice is to use uplink channel estimation as downlink channel estimation under time division multiplexing. This approach will encounter the impact of asymmetric transceiver radio frequency links in practice and bring phase noise. Phase noise, as a multiplicative noise, has a greater impact on the precoding effect and cannot be improved by increasing the signal-to-noise ratio. The academic community has conducted some research on distributed transmission and robust precoding. However, there are few effective solutions for robust distributed precoding suitable for non-cellular networks, and the non-idealities of the communication channels between the computing units of the fronthaul network (noise, quantization, etc.) may be encountered in real systems and affect the final performance. Summary of the invention

[0004] Purpose of the invention: The present invention provides a downlink robust distributed precoding implementation system and implementation method for a non-cellular wireless access network to solve the problems of high performance implementation of existing non-cellular large-scale MIMO distributed precoding and robustness to phase noise.

[0005] Technical solution: A downlink robust distributed precoding implementation system for a non-cellular wireless access network according to the present invention comprises: a first unit, a second unit, an Ethernet switching unit and a third unit for each frequency band;

[0006] The first unit completes the transmission and reception of wireless radio frequency signals in each frequency band, digital-to-analog and analog-to-digital conversion, and waveform signal processing of physical channels;

[0007] The second unit completes channel parameter estimation, downlink precoding calculation and precoding;

[0008] The third unit calculates downlink statistical precoding according to the statistical information sent by the second unit;

[0009] The Ethernet switching unit connects different units and enables information exchange.

[0010] Furthermore, the second unit is connected to one or more first units, the connection relationship between the second unit and the first unit is quasi-static, and the first unit reports all real-time channel state information to the corresponding second unit.

[0011] Furthermore, the second units are divided into several clusters, and the second units in the same cluster exchange instantaneous or statistical channel state information through direct connection or Ethernet switching units, but second units in different clusters do not need direct communication links.

[0012] Furthermore, the channel state information in the second unit is obtained in the channel estimation stage, and the first units connected to the second unit each upload the received pilot signal and the estimated phase noise statistical parameters, which are summarized by the second unit for channel estimation; the second unit calculates the local precoding using the estimated local channel, and the second units in the same cluster exchange information for calculating the second-level collaborative precoding, and the joint precoding within the cluster is achieved by cascading local precoding and collaborative precoding.

[0013] Furthermore, multiple third units exchange static or quasi-real-time channel state information through direct connection or Ethernet switching units; each third unit is directly connected to a cluster of second units or connected through an Ethernet switching unit, and the second units report static or quasi-real-time channel state information to the third unit to complete distributed precoding.

[0014] Furthermore, each third unit aggregates the channel statistical information and phase noise statistical parameters of the second unit to which it is directly connected, and sends the information through the communication link with other third units, calculates the collaborative precoding at the third unit level, and realizes the joint precoding of the entire cellular network by cascading the local precoding, collaborative precoding and collaborative precoding at the third unit level on the second unit.

[0015] Accordingly, a method for implementing downlink robust distributed precoding in a non-cellular wireless access network includes the following steps:

[0016] Step 1: Each first unit managed by each second unit receives a pilot signal sent by a user. Upload to the second unit and the received data signal Upload to the second unit. At the same time, the first unit also needs to upload the statistics of its phase noise to support robust precoding;

[0017] Step 2: The second unit receives the pilot signal obtained from the first unit, performs minimum mean square error channel estimation, and calculates precoding using the estimated channel;

[0018] Step 3: The third unit aggregates all the statistics of a cluster of second units directly connected to it, exchanges its own statistics with other third units, and finally the third unit adjusts the transmitted symbols through the exchanged statistical information to achieve joint transmission.

[0019] Furthermore, in Step 1, each first unit managed by each second unit will receive the pilot signal sent by the user Upload to the second unit and the received data signal Upload to the second unit. At the same time, the first unit also needs to upload the statistics of its phase noise to support robust precoding, which specifically includes the following steps:

[0020] Step 101: The first unit under the second unit under any third unit receives the assigned pilot signal sent by the user

[0021]

[0022] where c is the serial number of the third unit, m is the serial number of the second unit, l is the serial number of the first unit and l ∈ {1, 2, …, L}, k is the user sequence and k ∈ {1, 2, …, K}, represents the pilot sequence assigned to user k, τ p represents the pilot length, τ p <K means there is pilot contamination, N c(l represents independent complex Gaussian noise; each first unit will upload its own to the corresponding second unit m through the connection between the first unit and the second unit, which belongs to the channel estimation stage. The first unit is associated with at most τ p users, that is, a user-centered scalable cell-free network is realized;

[0023] Step 102: The first unit calculates the statistical parameters of the phase noise and transmits them to the corresponding second unit, including the mean value Assume independent and identically distributed, so that the second unit can know the statistical parameters of the phase noise of the first unit it manages.

[0024] Furthermore, in Step 2, the second unit receives the pilot signal obtained from the first unit, performs minimum mean square error channel estimation, and calculates precoding, which specifically includes the following steps:

[0025] Step 201: The second unit receives the pilot signal obtained from the first unit and performs channel estimation. In the minimum mean square error (MMSE) channel estimation, the estimated channel is expressed as

[0026]

[0027] in and The variance matrix of

[0028]

[0029] Estimation Error The variance of

[0030]

[0031] The channel estimate on the second unit is Then the channel estimate for each first unit is

[0032] Step 202: The second unit in each cluster performs precoding calculation using the estimated channel state information. The precoding matrix on each second unit is expressed as

[0033] W cm =B cm C cm

[0034] Among them B cm Indicates that only local channel information is used The local precoding of

[0035]

[0036] Among them A cm It consists of L×L matrix blocks, each of which Represented as an N×N matrix

[0037]

[0038] Where P .1 I is the normalized noise variance matrix, C cm is obtained as described in step 203;

[0039] Step 203: The second units under the management of the same third unit c exchange instantaneous or statistical joint transmission auxiliary information I XYZ , I XYZ is the local equivalent channel H \,cm Or the statistics of the local equivalent channel H \,cm Calculated by the following formula:

[0040]

[0041] The statistical equivalent channel is replaced by a numerical average. The number of second units that send instantaneous local equivalent channels accounts for no more than a set scalar ρ of the total number of second units in the cluster. Each second unit calculates the information it shares with other second units and sends it to other members in the cluster. Each second unit adjusts the local precoding through the auxiliary information obtained through the exchange. Then, C in step 202 cm It is obtained by the following c-th linear system:

[0042]

[0043] in It is the subscript of the shared instant in the second unit managed by the third unit c, Correspondingly The complement of is the phase noise that is statistically averaged out.

[0044] Further, in step 3, the third unit aggregates all statistics of a cluster of second units directly connected thereto, and exchanges its own statistics with other third units, and finally the third unit adjusts the transmission symbol through the exchanged statistical information to realize joint transmission, specifically including the following steps:

[0045] Step 301: The third unit generates statistical information on each second unit in the second unit cluster directly connected to the third unit. Sum, where In step 203 C cm The statistical average of the results is then Share with other third units;

[0046] Step 302: The third unit adjusts the transmitted symbol using the statistics obtained in step S01.

[0047] s c =D c s

[0048] Where D c It is solved by the following linear system:

[0049]

[0050] The entire precoding is expressed as B cm C cm D c That is, a layered structure.

[0051] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages: the robust cascade team precoding proposed in the present invention can implement robust precoding in a distributed manner, and can control the fronthaul capacity, achieving the best compromise between performance and cost; in addition, the robust cascade team precoding can also take into account a variety of practical problems, including but not limited to phase noise, and is a universal distributed robust precoding framework. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 It is a schematic diagram of the structure of the implementation system of the present invention.

[0053] Figure 2 The figure is a flow chart of the implementation method of the present invention.

[0054] Figure 3 This is a Monte Carlo simulation result diagram of the robust distributed precoding of the present invention, comparing the cumulative distribution of user rates of fully centralized and fully distributed uplinks.

[0055] Figure 4 This is a Monte Carlo simulation result diagram of the robust distributed precoding of the present invention, comparing the cumulative distribution of user rates in the case of no phase noise and the case of non-robust precoding with phase noise. DETAILED DESCRIPTION

[0056] A downlink robust distributed precoding implementation system for a non-cellular wireless access network is disclosed to achieve joint transmission and robustness against channel phase errors, and to realize a hierarchical precoding structure and flexible fronthaul network exchange cost control. Different from the traditional compromise method for large-scale fading decoding, the method of the present invention can achieve a fully centralized effect.

[0057] For the sake of convenience, it is stipulated that K represents the number of users, N represents the number of antennas of each first unit, L represents the number of first units connected to each second unit, M represents the number of second units managed by each third unit, C represents the number of third units, and each second unit is associated with one and only one third unit. Specifically, Figure 1 As shown, the connection between the second unit and the third unit can be realized through the switch, and the second units connected to the same third unit can interact through the switch. Similarly, the third units can exchange information through the switch. The channel between the first unit and the user is assumed to be a spatially correlated Rayleigh channel

[0058]

[0059] and is the large-scale fading from the first unit l to user k, where cmlk are the third unit number, the second unit number, the first unit number, and the user number in sequence. For the downlink channel, if phase noise is considered, it should be multiplied point by point by the noise vector f introduced due to the channel phase error cml , it is usually considered that the phase noise is the same within the same first unit, but different and independent between different first units.

[0060] The implementation of the system of the present invention can be divided into three parts, and each part has several:

[0061] 100 is the first unit;

[0062] 200 is the second unit;

[0063] 300 is the third unit, usually a Central Processing Unit (CPU).

[0064] In this embodiment, the robust distributed precoding of the cell-free network is divided into three steps:

[0065] Step S1. Each first unit managed by each second unit uploads the pilot signal sent by the user it receives to the second unit, and uploads the received data signal to the second unit. At the same time, the first unit also needs to upload the statistics of its phase noise to support robust precoding.

[0066] The specific steps are as follows:

[0067] Step S101. The first unit under the second unit under any third unit receives the allocated pilot signal sent by the user

[0068]

[0069] where c is the third unit number, m is the second unit number, l is the first unit number and l ∈ {1, 2,..., L}, k is the user sequence and k ∈ {1, 2,..., K}, represents the pilot sequence allocated to user k, τ p represents the pilot length (usually τ p < K, that is, there is pilot contamination), N c(l represents independent complex Gaussian noise. Each first unit uploads its own to the corresponding second unit m through the connection between the first unit and the second unit. This process belongs to the channel estimation stage, and the second unit in step S2 actually completes the channel estimation. Optionally, the first unit is connected to at most τ pThe user association is implemented, that is, a user-centric scalable non-cellular network is realized, and the selection method of the first unit and the user association includes but is not limited to distance-based, large-scale fading-based and other analytical or heuristic algorithms.

[0070] Step S102: The first unit calculates the statistical parameters of the phase noise and transmits them to the corresponding second unit, including but not limited to the mean (For ease of representation, assume independent and identical distribution), so that the second unit can know the statistical parameters of the phase noise of the first unit it manages.

[0071] Step S2: The second unit receives the pilot signal obtained from the first unit, performs channel estimation including but not limited to minimum mean square error, and uses the estimated channel to calculate precoding. Typically, local minimum mean square error (LMMSE) precoding can be used.

[0072] The specific steps include the following:

[0073] Step S201: The second unit receives the pilot signal obtained from the first unit and performs channel estimation. Taking the minimum mean square error (MMSE) channel estimation as an example, the estimated channel can be expressed as

[0074]

[0075] in and The variance matrix of

[0076]

[0077] Furthermore, the estimated error The variance of can be expressed as follows.

[0078]

[0079] The channel estimate on the second unit is Then the channel estimate for each first unit is

[0080] Step S202: The second unit in each cluster performs precoding calculation using the estimated channel state information. The precoding matrix on each second unit can be expressed as

[0081] E cm =B cm C cm

[0082] Among them B cm Indicates that only local channel information is used The local precoding can be expressed as

[0083]

[0084] Among them A cm It consists of L×L matrix blocks, each of which It can be represented as an N×N matrix

[0085]

[0086] Where P .1 I is the normalized noise variance matrix. C cm It is obtained by the step S203.

[0087] Step S203: The second units under the management of the same third unit c exchange instantaneous or statistical joint transmission auxiliary information I XYZ In this embodiment, I XYZ is the local equivalent channel H \,cm Or the statistics of the local equivalent channel H \,cm Calculated by the following formula:

[0088]

[0089] The statistical equivalent channel can be replaced by a statistic or, optionally, by a numerical average. The number of second units that send instantaneous local equivalent channels accounts for no more than a set scalar ρ of the total number of second units in the cluster. Each second unit calculates the information it shares with other second units and sends it to other members in the cluster. Each second unit adjusts the local precoding based on the auxiliary information obtained through the exchange. Then, C in step S202 cm It is obtained by the following c-th linear system:

[0090]

[0091] in It is the subscript of the shared instant in the second unit managed by the third unit c, Correspondingly The complement of is the phase noise that is statistically averaged out.

[0092] Step S3: The third unit aggregates all statistics of a cluster of second units to which it is directly connected, and exchanges its own statistics with other third units. Finally, the third unit adjusts the transmission symbol based on the exchanged statistical information to realize joint transmission. Optionally,

[0093] The specific steps include the following:

[0094] Step S301: The third unit generates statistical information on each second unit in the second unit cluster directly connected to the third unit. Sum, where In step S203 C cm The statistical average of the results is then Share with other third units.

[0095] Step S302: The third unit adjusts the transmitted symbol using the statistics obtained by the exchange in step S301.

[0096] s c =D c s

[0097] Where D c It can be obtained by solving the following linear system:

[0098]

[0099] Then the entire precoding can be expressed as B cm C cm D c That is, a stacked structure. All the above solution formulas can be proved to be correct through team decision theory.

[0100] Figure 3 The method proposed by the present invention ( Figure 3 The middle figure is a curve of “partial Joint Team MMSE”, and “rho” represents ρ in step S203 . It can be seen that this method can achieve a smooth compromise between performance and the cost of exchanged information volume. Figure 4 The phase noise robustness of the method of the present invention is demonstrated, where "JMMSE" represents the precoding performance in the absence of phase noise and "PhasedJTMMSE" represents the performance of non-robust precoding in the presence of phase noise, which obviously has an improvement of more than 20% at 95% user rate.

Claims

1. A downlink robust distributed precoding implementation system for a non-cellular wireless access network, characterized in that: include: The first unit, the second unit, the Ethernet switching unit and the third unit of each frequency band; The first unit completes the transmission and reception of wireless radio frequency signals in each frequency band, digital-to-analog and analog-to-digital conversion, and waveform signal processing of physical channels; The second unit completes channel parameter estimation, downlink precoding calculation and precoding; The third unit calculates downlink statistical precoding according to the statistical information sent by the second unit; The Ethernet switching unit connects different units and enables information exchange.

2. The downlink robust distributed precoding implementation system of the non-cellular wireless access network according to claim 1, characterized in that: The second unit is connected to one or more first units, the connection relationship between the second unit and the first unit is quasi-static, and the first unit reports all real-time channel state information to the corresponding second unit.

3. The downlink robust distributed precoding implementation system of the non-cellular wireless access network according to claim 1, characterized in that: The second units are divided into several clusters. The second units in the same cluster exchange instantaneous or statistical channel state information through direct connection or Ethernet switching units, but second units in different clusters do not need direct communication links.

4. The downlink robust distributed precoding implementation system of the non-cellular wireless access network according to claim 1, characterized in that: The channel state information in the second unit is obtained in the channel estimation stage. The first units connected to the second unit each upload the received pilot signal and the estimated phase noise statistical parameters, which are summarized by the second unit for channel estimation. The second unit calculates the local precoding using the estimated local channel. The second units in the same cluster exchange information for calculating the second-level collaborative precoding, and the joint precoding within the cluster is achieved by cascading local precoding and collaborative precoding.

5. The downlink robust distributed precoding implementation system of the non-cellular wireless access network as claimed in claim 1, characterized in that: Multiple third units exchange static or quasi-real-time channel state information through direct connection or Ethernet switching unit; each third unit is directly connected to a cluster of second units or connected through Ethernet switching unit, and the second unit reports static or quasi-real-time channel state information to the third unit to complete distributed precoding.

6. The downlink robust distributed precoding implementation system of the non-cellular wireless access network according to claim 1, characterized in that: Each third unit aggregates the channel statistical information and phase noise statistical parameters of the second unit to which it is directly connected, and sends the information through the communication link with other third units, calculates the collaborative precoding at the third unit level, and realizes the joint precoding of the entire cellular network by cascading the local precoding on the second unit, collaborative precoding and collaborative precoding at the third unit level.

7. A method for implementing the downlink robust distributed precoding implementation system for the non-cellular wireless access network as claimed in claim 1, characterized in that: The steps include: Step 1: Each first unit managed by each second unit receives a pilot signal sent by a user. Upload to the second unit and receive the data signal Upload to the second unit, while the first unit also uploads its phase noise statistics to support robust precoding; Step 2: The second unit receives the pilot signal obtained from the first unit, performs minimum mean square error channel estimation, and uses the estimated channel to calculate precoding; Step 3: The third unit aggregates all statistics of a cluster of second units to which it is directly connected, and exchanges its own statistics with other third units. Finally, the third unit adjusts the transmission symbols based on the exchanged statistical information to achieve joint transmission.

8. The method for implementing downlink robust distributed precoding in a non-cellular wireless access network as claimed in claim 7, characterized in that: In step 1, each first unit managed by each second unit receives a pilot signal sent by a user Upload to the second unit and receive the data signal Upload to the second unit, and the first unit also uploads its phase noise statistics to support robust precoding. The specific steps include: Step 101: The first unit under the second unit under any third unit receives the assigned pilot signal sent by the user. where c is the serial number of the third unit, m is the serial number of the second unit, l is the serial number of the first unit and l ∈ {1, 2,..., L}, k is the user sequence and k ∈ {1, 2,..., K}, represents the pilot sequence assigned to user k, τ p represents the pilot length, τ p <K means there is pilot contamination, N cgl represents independent complex Gaussian noise; each first unit will upload its own to the corresponding second unit m through the connection between the first unit and the second unit, which belongs to the channel estimation stage. The first unit is associated with at most τ p users, that is, a scalable cell-free network centered on users is realized; Step 102: The first unit calculates the statistical parameters of the phase noise and transmits them to the corresponding second unit, including the mean Assuming independent and identical distributions, the second unit thus knows the statistical parameters of the phase noise of the first unit it manages.

9. The method for implementing downlink robust distributed precoding in a non-cellular wireless access network as claimed in claim 7, characterized in that: In step 2, the second unit receives the pilot signal obtained from the first unit, performs minimum mean square error channel estimation, and uses the estimated channel to calculate precoding, which specifically includes the following steps: Step 201: The second unit receives the pilot signal obtained from the first unit and performs channel estimation. In the minimum mean square error (MMSE) channel estimation, the estimated channel is expressed as in and The variance matrix of Estimation Error The variance of The channel estimate on the second unit is Then the channel estimate for each first unit is Step 202: The second unit in each cluster performs precoding calculation using the estimated channel state information. The precoding matrix on each second unit is expressed as W cm =B cm C cm Among them B cm Indicates that only local channel information is used The local precoding of Among them A cm It consists of L×L matrix blocks, each of which Represented as an N×N matrix Where P -1 I is the normalized noise variance matrix, C cm is obtained as described in step 203; Step 203: The second units under the management of the same third unit c exchange instantaneous or statistical joint transmission auxiliary information I aux , I aux is the local equivalent channel H e，cm Or the statistics of the local equivalent channel H e，cm Calculated by the following formula: The statistical equivalent channel is replaced by a numerical average. The number of second units that send instantaneous local equivalent channels accounts for no more than a set scalar ρ of the total number of second units in the cluster. Each second unit calculates the information it shares with other second units and sends it to other members in the cluster. Each second unit adjusts the local precoding through the auxiliary information obtained through the exchange. Then, C in step 202 cm It is obtained by the following c-th linear system: in It is the subscript of the shared instant in the second unit managed by the third unit c, Correspondingly The complement of is the phase noise that is statistically averaged out.

10. The method for implementing downlink robust distributed precoding in a non-cellular wireless access network according to claim 7, characterized in that: In step 3, the third unit aggregates all statistics of a cluster of second units directly connected to it, and exchanges its own statistics with other third units. Finally, the third unit adjusts the transmission symbol through the exchanged statistical information to realize joint transmission. Specifically, the steps include: Step 301: The third unit generates statistical information on each second unit in the second unit cluster directly connected to the third unit. Sum, where In step 203 C cm The statistical average of the results is then Share with other third units; Step 302: The third unit adjusts the transmitted symbol using the statistics obtained in step S01. s c =D c s Where D c It is solved by the following linear system: The entire precoding is expressed as B cm C cm D c That is, a layered structure.

Images