A channel estimation and pilot allocation method for pilot contamination elimination

Abstract

The present invention discloses a channel estimation and pilot allocation method for eliminating pilot pollution, which belongs to the field of wireless communications. The method of the present invention comprises: a central processing unit selects the AP channel most seriously polluted by pilots as the desired AP channel, calculates the overlap value of each interfering AP channel and the desired AP channel in sequence, allocates orthogonal pilot sequences to APs with high overlap, and allocates non-orthogonal pilot sequences to the rest; the desired AP performs channel estimation based on the non-overlapping of the delay-Doppler-angle domain channel subspace, and eliminates pilot pollution with the assistance of subspace structured covariance. The present invention has a more universal desired AP channel selection system and a more efficient and accurate pilot pollution elimination capability, improves the accuracy of judging the most seriously polluted AP or terminal, improves the pilot pollution elimination performance, realizes a more efficient pilot allocation scheme, improves the robustness of the system, and can be applied to systems composed of larger-scale APs.

Description

Technical Field

[0001] The present invention belongs to the field of wireless communications, and in particular relates to a channel estimation and pilot allocation method for eliminating pilot pollution. Background Art

[0002] Pilot contamination elimination is a method of avoiding the superposition of non-orthogonal pilot interference by rationally allocating orthogonal resources. Channel estimation is typically performed in communication systems using orthogonal pilots to differentiate the channel estimation results for different antenna ports and users. When pilot resources are insufficient, non-orthogonal pilots may appear for different antenna ports or users, making it impossible to differentiate their respective channel estimation results, thus reducing the performance of the communication system. This is known as pilot contamination. The basic idea behind pilot contamination elimination technology is to rationally allocate pilot resources by dividing sectors, utilizing the multiplexing gain of multiple sectors to prevent non-orthogonal pilots from appearing in the same sector, thereby eliminating pilot contamination interference. Another approach is to pre-compensate for pilot contamination, estimate the contamination, and restore pure channel state information.

[0003] Channel estimation is a core technology in communication systems. Wireless channels significantly impact system performance. Unlike wired channels, which are fixed, wireless channels exhibit significant uncertainty. Furthermore, wireless channels are significantly impacted by environmental factors, such as large-scale fading influenced by location and distance, time-varying channel characteristics, and frequency selectivity caused by multipath. Furthermore, the maturing of connected vehicle and airborne technologies requires wireless communication systems to account for user mobility. This exacerbates the time-varying characteristics of wireless channels and Doppler shift, resulting in additional frequency or phase offsets in transmitted signals. These factors combine to create a complex channel path between receiver and transmitter, often characterized by significant randomness. This poses significant challenges to receiver design. To optimize signal reception and detection at the receiver, wireless channel estimation is necessary. The accuracy of the channel estimate directly impacts system performance, making channel estimation a crucial technology in wireless communication systems.

[0004] In distributed massive MIMO (Multiple Input, Multiple Output) antenna systems, due to the complexity of the channel environment, high-precision, low-overhead channel estimation solutions are crucial for stable communication system performance. However, current research on channel estimation technology in distributed massive MIMO systems has not yet fully realized the system's potential, and pilot contamination issues still exist.

[0005] The invention patent application with publication number CN115174022A disclosed a pilot allocation method, apparatus, device and storage medium on October 11, 2022. The technical solution includes: determining the user equipment that needs to be pilot allocated and the pilot to be allocated in the communication system, and determining the preliminary pilot allocation scheme; calculating the worst-performing user equipment in the absence of pilot pollution based on the difference between the arrival angle of the interfering user equipment and the arrival angle of the desired user equipment; optimizing the preliminary pilot allocation scheme based on the first channel estimation results of all user equipment in the communication system and the second channel estimation results of the worst-performing user equipment in the absence of pilot pollution to obtain the optimal pilot allocation scheme. This technical solution not only improves the overall performance of the system but also ensures the performance of the worst user equipment, and can alleviate the impact of pilot pollution on large-scale antenna systems. However, this technical solution has the following disadvantages:

[0006] (1) The existing technology calculates the arrival angle difference of different users and determines the user equipment with the worst performance in the absence of pilot pollution based on the minimum sum of the angle differences. However, this determination method cannot completely determine the user equipment with the worst performance in the system because when two users overlap in arrival angle, they are both calculated as the user equipment with the worst performance according to the objective function, which is unreasonable.

[0007] (2) Existing technical solutions rely on the arrival angles of different users to distinguish channel state information. Using an objective function, they calculate the user with the least angular impact and assign non-orthogonal pilots. Furthermore, they assign orthogonal pilots to the user with the worst performance when there is no pilot contamination, thereby achieving better system performance. However, when the system has a large number of antennas or users, angle overlap may occur. In more complex systems, it is difficult to distinguish antenna ports using angle alone, and pilot contamination cannot be eliminated. Summary of the Invention

[0008] To address the shortcomings of existing technologies that use angular domain information between multiple wireless access points (APs) to distinguish different AP channels and then use this angular domain information for channel estimation and pilot contamination elimination, the present invention provides a channel estimation and pilot allocation method for pilot contamination elimination. This method uses the location information of distributed wireless access points (APs) within a cell system, including distance (corresponding to the delay domain) and angle (corresponding to the Doppler and angle domains), to distinguish different antenna ports or terminals under the same non-orthogonal pilot interference. By designing an overlap verification function, the overlap of channel support sets between different APs in the delay-Doppler-angle domain is determined, and pilot allocation is then performed.

[0009] The present invention provides a channel estimation and pilot allocation method for eliminating pilot contamination, comprising the following steps:

[0010] (1) The central processing unit performs pilot allocation based on wireless access point AP overlap verification, including:

[0011] (1.1) Calculate the mean square error (MSE) of the channel estimation for each AP under pilot contamination. Select the AP channel with the most severe pilot contamination as the desired AP channel, and the remaining AP channels as interfering AP channels. Assign independent orthogonal pilots to the desired AP channel.

[0012] (1.2) Traverse all interfering AP channels, calculate the overlap verification function value of each interfering AP channel and the desired AP channel in turn, and record the overlap verification function calculation results;

[0013] Overlap verification function between the bth AP channel and the expected AP channel b* as follows:

[0014]

[0015] in, b are the delay domain channel support sets of the desired AP channel and the bth AP channel, k b are the Doppler domain channel support sets of the desired AP channel and the bth AP channel, r b are the angle domain channel support sets of the desired AP channel and the b-th AP channel respectively; δ(·) represents the Dirac impulse function; and According to the function Υ N (x) calculation, N k Indicates the number of Doppler domain resource blocks, N T Indicates the number of antennas deployed on each AP;

[0016] (1.3) Sort the calculation results of the overlap verification function of all interfering AP channels and the desired AP channels in descending order, and S -1) calculation results corresponding to the AP assigned an orthogonal pilot sequence, the remaining APs are assigned non-orthogonal pilot sequences; where N S The length of the pilot sequence sent to the user;

[0017] (1.4) Allocating the pilot sequence allocated to each AP to the pilot sequence matrix Φ;

[0018] (2) The AP is expected to perform channel estimation based on the non-overlapping nature of the channel subspaces in the Delay-Doppler-Angle (DDA) domain and to eliminate pilot contamination using subspace structured covariance.

[0019] (2.1) After each AP antenna receives the signal from the end user, it calculates the time-frequency-space domain channel hTFS ;

[0020] (2.2) h TFS Transform to the delay-Doppler-angle domain, let h TFS After transformation, we get h DDA ;

[0021] (2.3) Obtain the delay-Doppler-angle domain channel support set based on the system prior information or the previous channel estimation result;

[0022] (2.4) in h DDA Traverse the channel support set obtained in step 2.3, and select the angular channel support set by the threshold of the angular channel

[0023] (2.5) Traverse the angular domain channel support set obtained in step 2.4 to obtain the delay-Doppler domain channel support set corresponding to each element of the angular domain channel support set

[0024] (2.6) in h DDA Based on the delay-Doppler domain channel support set obtained in step 2.5 Search for the angular region corresponding to the desired AP channel The delay-Doppler domain channel vector Then, a rough estimate of the desired AP channel is obtained as follows:

[0025]

[0026] in, is the expected AP channel estimation result obtained last time; the superscript T indicates transposition;

[0027] (2.7) Calculate the covariance matrix of the desired AP channel and obtain the subspace eigenvalue matrix and the eigenvector array

[0028] (2.8) Calculate the final channel estimation result of the desired AP channel after covariance filtering as follows:

[0029]

[0030] Among them, σ 2 represents the channel noise variance, express dimensional identity matrix, yes The rank of , w represents the noise.

[0031] The advantages and positive effects of the present invention are:

[0032] (1) The method of the present invention allocates limited orthogonal pilot resources to APs with high channel support set overlap, eliminating pilot contamination caused by channel support set overlap. The method of the present invention utilizes more comprehensive spatial information about APs, including range and angle domains, to distinguish antenna ports or terminals, and corresponds to three-dimensional channels in the delay-Doppler-angle domain. This provides a more universal desired AP channel selection system and more efficient and accurate pilot contamination elimination capabilities.

[0033] (2) The proposed subspace structured covariance-assisted channel estimation scheme, based on the proposed pilot allocation scheme, further utilizes the subspace non-overlapping property of the channel support set to improve the accuracy of channel estimation with non-orthogonal pilots. Compared with channel estimation schemes based solely on angular domain differentiation, the proposed scheme improves the accuracy of determining the most polluted AP or terminal, improves pilot contamination elimination performance, achieves a more efficient pilot allocation scheme, improves system robustness, and is applicable to systems with larger APs. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 Schematic diagram of the architecture of a multi-AP pilot pollution elimination system used by the method of the present invention;

[0035] Figure 2 It is a flow chart of the pilot allocation scheme implemented by the method of the present invention;

[0036] Figure 3 It is a flow chart of the method of the present invention based on subspace structured covariance assisted channel estimation;

[0037] Figure 4 This is an example diagram of the effect of using the method of the present invention to eliminate pilot pollution and thus restore the DDA domain channel. DETAILED DESCRIPTION

[0038] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0039] The present invention proposes a channel estimation and pilot allocation method for eliminating pilot contamination, which is applied in a scenario such as Figure 1 As shown, the overall architecture for the method of the present invention includes a central processing unit (CPU), access points (APs), and user terminals. The AP can be a base station or a specially placed dedicated relay device, deploying a linear antenna array. The terminal can be any access device, including but not limited to mobile phones, computers, and smart tablets. It can also be deployed on highly mobile vehicles such as high-speed trains, unmanned vehicles, and drones. The CPU is the baseband signal processing device in the system, responsible for calculating the AP received signal, performing channel estimation, and receiving detection.

[0040] The method of the present invention mainly includes two parts: a pilot allocation scheme based on AP overlap verification and a channel estimation scheme based on subspace structured covariance assistance. These two parts are described below.

[0041] In the prior art, only the different arrival angles between APs or terminals are considered to distinguish AP channels. The present invention considers the situation of using APs to deploy linear antenna arrays, deeply explores the geographical location factors of distributed multiple APs in the system, and uses the AP channel with the worst MSE performance under pilot pollution as the basis for judging the desired AP channel. A pilot allocation scheme based on delay-Doppler-angle domain channel overlap is designed, which decouples APs with higher channel support set overlap through pilot orthogonality distinction, and decouples APs with lower channel support set overlap through natural non-overlapping distinction, that is, allocating orthogonal pilot resources to APs with higher overlapping pilot pollution with the desired AP channel, thereby reducing the pilot pollution interference of the system. Figure 2 As shown in FIG, each AP has non-overlapping delay-Doppler-angle domain channel support sets due to its different geographical locations. Their pilot allocation schemes are determined by the overlap verification function. The specific steps for implementing the pilot allocation scheme are as follows.

[0042] Step 1.1: First, calculate the mean square error (MSE) performance of the channel estimation of each AP under pilot contamination. Select the AP channel with the most serious pilot contamination as the desired AP channel, and the remaining AP channels as interference AP channels.

[0043] Assume that the length of the pilot sequence sent by the user is N S , there are B APs in total, each AP has N T A linear array of antennas, when N S When <B, the pilot signals of all APs cannot be recovered based on the correlation operation, thus causing pilot contamination.

[0044] Define b* to represent the desired AP channel, and the index of other APs is b. represents the mean square error of the b-th AP channel estimation.

[0045] in, Defined as the b-th AP channel covariance matrix, h b represents the bth AP channel, σ 2 represents the channel noise variance, I N Denotes the N-dimensional identity matrix, tr denotes the mean square error, and the superscript H denotes the conjugate transpose of the matrix. After selecting the desired AP channel, an independent orthogonal pilot is assigned to it.

[0046] Step 1.2, traverse all other interfering APs, and calculate the overlap verification function value of each interfering AP channel and the expected AP channel in turn, and record the overlap verification function calculation results. The overlap verification function of the bth AP channel and the expected AP channel is defined as as follows:

[0047]

[0048] in, b are the channel support sets of the expected AP channel and the bth AP channel in the delay domain, k b are the channel support sets of the desired AP channel and the bth AP channel in the Doppler domain, r b are the channel support sets of the desired AP channel and the bth AP channel in the angular domain respectively; δ(·) represents the Dirac impulse function; and According to the function Υ N (x) calculation, N k Indicates the number of Doppler domain resource blocks, N T Indicates the number of antennas deployed on the AP. Function Y N (x) is defined as follows:

[0049]

[0050] Where, j represents the imaginary unit; x is the input data; N is the accumulated sum number; It means "defined as".

[0051] The channel support set represents the set of indices of the three-dimensional channel matrix locations where non-zero channel values ​​are located. The channel support set is composed of subsets of the delay domain, Doppler domain, and angle domain channel support sets. The channel support set is obtained based on system prior information or the previous channel estimation results.

[0052] Step 1.3, sort the calculation results of the overlap verification function of all interfering AP channels and the desired AP channels in descending order, and calculate the length of the preamble sequence in the queue sequence minus one (N S -1) APs with the desired AP channel overlap are assigned orthogonal pilot sequences, while the remaining APs with low channel overlap are assigned non-orthogonal pilot sequences. This reduces the impact of pilot contamination by using the non-overlapping subspace of the channel support set, thus eliminating system pilot contamination.

[0053] like Figure 2As shown, it is determined whether there is an unconfigured orthogonal pilot sequence. If not, the allocation is terminated. If so, it is determined whether the overlap between the current AP channel and the expected AP channel is the largest. If so, an orthogonal pilot sequence is allocated. If not, a non-orthogonal pilot sequence is allocated. The overlap corresponding to the APs allocated with the pilot sequence is set to 0, and then the pilot sequence allocation process is repeated until the pilot sequences of all APs are allocated.

[0054] Step 1.4: Configure the selected pilot sequence resources allocated to each AP into the pilot matrix as follows:

[0055]

[0056] in represents the b-th segment pilot sequence vector, that is, the pilot sequence allocated to the b-th AP, The pilot sequence matrix that finally meets the pilot allocation scheme is obtained.

[0057] In the prior art, AP channel estimation under pilot contamination is directly performed through coarse estimation in the angular domain. However, the present invention considers, based on the pilot allocation scheme proposed above, utilizing the spatial position characteristics of the system's distributed APs, namely the distance domain (delay domain) and the angle domain (Doppler domain and angle domain), first performing a conversion domain operation for coarse channel estimation. Orthogonal pilots are used for channel estimation in AP channels with high overlap in the delay-Doppler-angle domain channel support sets, and orthogonal sequence code division multiplexing is performed at the receiving end. A subspace structured covariance-assisted scheme is used in AP channels with low overlap in the channel support sets. Through eigenvalue decomposition of the covariance, due to the low overlap in the channel support sets, the eigenvectors fall into different subspaces. By setting a reasonable size of the delay-Doppler-angle domain data resource blocks, pilot contamination can be completely eliminated, thereby improving the channel estimation accuracy through covariance-assisted coarse estimation filtering.

[0058] like Figure 3 As shown, after determining the pilot allocation scheme for each AP, the AP performs channel estimation based on the non-overlapping nature of the delay-Doppler-angle domain channel subspaces. This is aided by subspace structured covariance, which effectively eliminates pilot contamination. The present invention implements the following steps for a subspace structured covariance-assisted channel estimation scheme for pilot contamination elimination of the desired AP channel.

[0059] In step 2.1, after each AP antenna receives the signal from the end user, it calculates the time-frequency-space (TFS) channel. Each AP antenna receives a time-frequency-space channel that is contaminated by the pilot signal, which is expressed as: where h TFS represents the channel vector in TFS domain, represents the pseudo-inverse of the matrix, and y represents the received signal vector.

[0060] Step 2.2, calculate the time-frequency-space domain channel h TFS Perform transform domain processing and transform it into the delay-Doppler-angle domain. Let h TFS Transform the domain to get h DDA .

[0061] In step 2.3, the delay-Doppler-angle domain channel support set is obtained based on the system's prior information or the previous channel estimation results. The system's prior information records the spatial position of each AP's channel. Therefore, during the coherence time, each AP is assumed to be a quasi-stationary channel. The channel support sets for different APs can be directly obtained from the system's internal AP database. Alternatively, the channel support set for the current channel estimate can be obtained by obtaining the index set of the non-zero channel value locations based on the previous channel estimation results.

[0062] Determine whether the accuracy of the last channel estimation obtained meets the requirements. If so, use the last channel estimation result to obtain the channel support set. Otherwise, use the system prior information to obtain the channel support set.

[0063] Assume that the delay-Doppler-angle domain channel support set of the desired AP channel is expressed as

[0064] Step 2.4, in h DDA Traverse the channel support set in step 2.3 and select the angular domain channel support set by judging the threshold of the angular domain channel.

[0065] By fixing the index of the delay domain and Doppler domain, the angular domain channel support set of the desired AP channel is selected

[0066] Step 2.5, traverse the angular domain channel support set in step 2.4 to obtain the delay-Doppler domain channel support set corresponding to each element of the angular domain channel support set.

[0067] Assume that the delay-Doppler domain channel support set of the desired AP channel is obtained in Indicates fixed The delay-Doppler domain channel support set obtained when

[0068] In step 2.6, the delay-Doppler domain channel is searched and a rough estimate of the channel is obtained based on the support set of step 2.5.

[0069] Rough estimation result of desired AP channel in, is the expected AP channel estimation result obtained last time, is the rough estimation result of the expected AP channel for this update; is based on Search h DDA Get the angular domain corresponding to the desired AP channel The delay-Doppler domain channel vector of .

[0070] In step 2.7, the covariance matrix of the AP channel, as well as the subspace eigenvalues ​​and eigenvectors are calculated.

[0071] The covariance matrix of the bth AP channel And calculate the subspace eigenvalues ​​and eigenvectors: in is the eigenvector matrix, n b R b rank, is the eigenvalue matrix. The superscript H indicates the conjugate transpose, Indicates the bth AP channel, the delay domain index corresponding to this channel is l b , Doppler domain index is k b , angle domain index is r b .

[0072] Similarly, the eigenvector matrix of the desired AP channel can be calculated and the eigenvalue matrix

[0073] Step 2.8, based on the results of the above steps, calculate the final covariance filtered channel estimate.

[0074] For the desired AP channel, the channel estimation result obtained after covariance filtering is as follows:

[0075]

[0076] in, express dimensional identity matrix, yes The rank of , w represents noise, and the superscript H represents transposed conjugate.

[0077] Experiments were conducted on the method of the present invention. Experimental scenario: There are 20 APs scattered in a circular area with a distance of 250m centered on the user to serve the user. The pilot sequence length is 10, so 10 APs will use the same pilot sequence, which is subject to pilot contamination. The uplink channel model from the user to each AP uses the channel model containing 6 main paths in the 3GPP standard, the carrier center frequency is 4.9GHz, the number of AP antennas is 64, and the number of subcarriers and symbols in the orthogonal frequency division multiplexing (OFDM) resource block are 1024 and 128 respectively. The method of the present invention is used to perform channel estimation and pilot allocation for pilot contamination elimination, and the obtained DDA domain channel recovery is as follows: Figure 4 As shown in the figure, (a) and (b) demonstrate the decoupling of TFS interference superposition channels through the Delay-Doppler (DD) domain subspace, corresponding to the non-equidistant case in the pilot allocation scheme. Asymptotic orthogonality of the channel in the delay domain is achieved through the potential range-domain gain of the distributed cell-free massive MIMO system. (c) and (d) demonstrate the elimination of pilot contamination through the angle-domain subspace, corresponding to the equidistant case in the pilot allocation scheme. Asymptotic orthogonality of the channel in the angle domain is achieved through the spatial-domain gain of the distributed cell-free massive MIMO system. Experiments demonstrate that through the rational pilot allocation method of the present invention, ideal orthogonality can be achieved even with limited total resource blocks, achieving excellent pilot contamination elimination performance.

[0078] Except for the technical features described in the specification, all other technical features are known to those skilled in the art. The present invention omits descriptions of well-known components and well-known technologies to avoid redundancy and unnecessary limitation of the present invention. The implementation methods described in the above embodiments do not represent all implementation methods consistent with the present application. Based on the technical solution of the present invention, various modifications or variations that can be made by those skilled in the art without creative effort are still within the scope of protection of the present invention.

Claims

1. A channel estimation and pilot allocation method for pilot contamination elimination, characterized in that: The steps include: (1) The central processing unit performs pilot allocation based on AP overlap verification, including: (1.1) Calculate the mean square error (MSE) of the channel estimation for each AP under pilot contamination. Select the AP channel with the most severe pilot contamination as the desired AP channel, and the remaining AP channels as interfering AP channels. Assign independent orthogonal pilots to the desired AP channel. (1.2) Traverse all interfering AP channels, calculate the overlap verification function value of each interfering AP channel and the desired AP channel in turn, and record the overlap verification function calculation results; No. AP channel and expected AP channel Overlap verification function as follows: ； in, 、 are the delay domain channel support sets of the desired AP channel and the bth AP channel, 、 are the Doppler domain channel support sets of the desired AP channel and the bth AP channel, 、 are the angle domain channel support sets of the desired AP channel and the bth AP channel respectively; represents the Dirac impulse function; and According to the function Calculation, N k Indicates the number of Doppler domain resource blocks, N T Indicates the number of antennas deployed by each AP; the function The definition is as follows: ； Where j is the imaginary unit, x is the input data, and N is the accumulated summation number; (1.3) Sort the calculation results of the overlap verification function of all interfering AP channels and the desired AP channels in descending order, and -1) calculation results corresponding to the AP is assigned an orthogonal pilot sequence, and the remaining APs are assigned non-orthogonal pilot sequences; where, The length of the pilot sequence sent to the user; (1.4) Configure the pilot sequence allocated to each AP into the pilot sequence matrix ; (2) The AP is expected to perform channel estimation based on the non-overlapping nature of the delay-Doppler-angle domain channel subspaces and to eliminate pilot contamination with the aid of subspace structured covariance. (2.1) After each AP antenna receives the signal from the end user, it calculates the time-frequency-space domain channel ; (2.2) Transform to the delay-Doppler-angle domain, let After transformation, we get ; (2.3) Obtain the delay-Doppler-angle domain channel support set based on the system prior information or the previous channel estimation result; (2.4) Traverse the channel support set obtained in step 2.3, and select the angular channel support set by the threshold of the angular channel ; (2.5) Traverse the angular domain channel support set obtained in step 2.4 to obtain the delay-Doppler domain channel support set corresponding to each angular domain channel support set element ; (2.6) In Based on the delay-Doppler domain channel support set obtained in step 2.5 Search for the angular region corresponding to the desired AP channel The delay-Doppler domain channel vector , and then obtain the rough estimate of the desired AP channel as follows: ， in, is the expected AP channel estimation result obtained last time; the superscript T indicates transposition; (2.7) Calculate the covariance matrix of the desired AP channel and obtain the subspace eigenvalue matrix and the eigenvector array ; (2.8) Calculate the final channel estimation result of the desired AP channel after covariance filtering ,as follows: ； in, represents the channel noise variance, express dimensional identity matrix, yes The rank of , w represents noise, and the superscript H represents transposed conjugate.

2. The method according to claim 1, characterized in that In step 1.4, the pilot sequence assigned to the bth AP is expressed as , then the obtained pilot sequence matrix ; Among them, I N represents the N-dimensional identity matrix, represents the Kronecker product.

3. The method according to claim 1, characterized in that In the step 2.3, it is determined whether the accuracy of the channel estimation obtained last time meets the requirements. If so, the index set of the three-dimensional positions of the channel non-zero values ​​in the delay domain, Doppler domain, and angle domain is obtained using the last channel estimation result to obtain the delay-Doppler-angle domain channel support set; if not, the delay-Doppler-angle domain channel support set of the AP is obtained from each AP database based on the system prior information.

Images