Multiple-input and multiple-output detection device and method based on recursive tree search

Abstract

According to one aspect of the present disclosure, a baseband chip may be configured to receive a data stream associated with a channel. The baseband chip may be configured to select a first anchor point from a first constellation point of a first transmission layer. The baseband chip may be configured to select a first subset of constellation points from the first constellation point and a second constellation point. The baseband chip may be configured to perform a first iteration of a recursive tree search. The baseband chip may be configured to determine a first path metric based at least in part on the first iteration of the recursive tree search. The baseband chip may be configured to select a second anchor point from a second constellation point of a second transmission layer. The second anchor point may be associated with a second iteration of the recursive tree search.

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims priority to U.S. Provisional Patent Application No. 63 / 024,845, filed on May 14, 2020, entitled “METHOD OF RECURSIVE TREE SEARCHBASED MIMO DETECTION,” the entire contents of which are incorporated herein by reference. Technical Field

[0003] Embodiments of the present disclosure relate to apparatus and methods for wireless communications. Background Art

[0004] Wireless communication systems are widely deployed to provide a variety of telecommunication services, such as telephony, video, data, messaging, and broadcasting. In cellular communications, such as the fourth-generation (4G) Long Term Evolution (LTE) and the fifth-generation (5G) New Radio (NR), the 3rd Generation Partnership Project (3GPP) has defined various signal detection mechanisms, such as multiple-input multiple-output (MIMO) detection. Summary of the Invention

[0005] Embodiments of apparatus and methods for recursive tree search-based MIMO detection are disclosed herein.

[0006] According to another aspect of the present disclosure, a baseband chip including a memory and at least one processor coupled to the memory is configured to perform various operations. In some embodiments, the at least one processor is configured to receive a data stream associated with a channel. In certain aspects, the channel includes multiple transmission layers. In other certain aspects, the multiple transmission layers may include a first transmission layer associated with a first constellation point and a second transmission layer associated with a second constellation point. In some other embodiments, the at least one processor may be configured to select a first anchor point from the first constellation point of the first transmission layer. In some other embodiments, the at least one processor may be configured to select a first subset of constellation points from the first constellation point and the second constellation point based at least in part on the first anchor point. In some other embodiments, the at least one processor may be configured to perform a first iteration of a recursive tree search operation based at least in part on the first anchor point and the first subset of constellation points. In some other embodiments, the at least one processor may be configured to determine a first path metric / path based at least in part on the first iteration of the recursive tree search operation. In some other embodiments, the at least one processor may be configured to select a second anchor point from the second constellation point of the second transmission layer based at least in part on the first path metric / path. In particular aspects, the second anchor point may be associated with a second iteration of the recursive tree search operation.

[0007] According to one aspect of the present disclosure, a baseband chip including a memory and at least one processor coupled to the memory is configured to perform various operations. In some embodiments, the at least one processor is configured to receive a data stream associated with a channel. In certain aspects, the channel may include multiple transmission layers. In other certain aspects, the multiple transmission layers may include a first transmission layer and a second transmission layer associated with a resource block (RB). In certain aspects, the RB may include at least one reference resource element (RE) and a plurality of data REs. In some other embodiments, the at least one processor may be configured to perform an initial noise estimate of the channel based at least in part on the reference REs. In some other embodiments, the at least one processor is configured to perform a first noise whitening operation on the channel based at least in part on the initial noise estimate. In some other embodiments, the at least one processor is configured to perform a first iteration of a recursive tree search operation on each of the plurality of first data REs for the first RB to determine first symbol estimates associated with the first and second transmission layers. In some other embodiments, the at least one processor is configured to perform subsequent noise estimates of the channel based at least in part on the reference REs, the plurality of data REs, and the first symbol estimates associated with the first iteration.

[0008] According to yet another aspect of the present disclosure, a method is disclosed that may include receiving a data stream associated with a channel. In certain aspects, the channel may include multiple transmission layers. In other certain aspects, the multiple transmission layers may include a first transmission layer associated with a first constellation point and a second transmission layer associated with a second constellation point. In some other embodiments, the method may include selecting a first anchor point from the first constellation point of the first transmission layer. In some other embodiments, the method includes selecting a first subset of constellation points from the first constellation points based at least in part on the first anchor point. In some other embodiments, the method may include performing a first iteration of a recursive tree search operation based at least in part on the first anchor point and the first subset of constellation points. In some other embodiments, the method may include determining a first best path based on a path metric from the first iteration of the recursive tree search operation. In some other embodiments, the method may include selecting a second anchor point from the second constellation point of the second transmission layer based at least in part on the first best path. The second anchor point is associated with a second iteration of the recursive tree search operation.

[0009] According to another aspect of the present disclosure, a method is disclosed that may include receiving a data stream associated with a channel. In a specific aspect, the channel may include multiple transmission layers. In other specific aspects, the multiple transmission layers include a first transmission layer and a second transmission layer associated with an RB. In other specific aspects, the RB includes at least one reference RE and a plurality of data REs. In some embodiments, the method may also include performing an initial noise estimate of the channel based at least in part on the reference RE. In some embodiments, the method may also include performing a first noise whitening operation on the channel based at least in part on the initial noise estimate. In some embodiments, the method may also include, for the RB, performing a first iteration of a recursive tree search operation on each of the plurality of data REs to determine a first symbol estimate for the first and second transmission layers. In some embodiments, the method may also include performing a subsequent noise estimate of the channel based at least in part on the reference RE and the plurality of estimated data REs associated with the first iteration. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The accompanying drawings, which are incorporated herein and form a part of the specification, illustrate embodiments of the present disclosure and, together with the description, further serve to explain the principles of the present disclosure and enable one skilled in the relevant art to make and use the present disclosure.

[0011] Figure 1 An exemplary wireless network according to some embodiments of the present disclosure is shown.

[0012] Figure 2 A block diagram of a device including a baseband chip, a radio frequency (RF) chip, and a host chip according to some embodiments of the present disclosure is shown.

[0013] Figure 3 A block diagram of a first exemplary baseband chip according to some embodiments of the present disclosure is shown.

[0014] Figure 4 A block diagram of a second exemplary baseband chip according to some embodiments of the present disclosure is shown.

[0015] Figure 5 A flow chart of a first exemplary method for MIMO detection data processing based on recursive tree search according to some embodiments of the present disclosure is shown.

[0016] Figure 6 A flow chart of a second exemplary method for MIMO detection data processing based on recursive tree search according to some embodiments of the present disclosure is shown.

[0017] Figure 7 A block diagram of an exemplary node according to some embodiments of the present disclosure is shown.

[0018] Figure 8 A lattice structure that can be used in a traditional fixed-complexity tree search is shown.

[0019] Figure 9 Shown is a block diagram of a traditional baseband chip.

[0020] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. DETAILED DESCRIPTION

[0021] Although specific configurations and arrangements have been discussed, it should be understood that the above discussion is for illustrative purposes only. Those skilled in the relevant art will recognize that other configurations and arrangements may be used without departing from the spirit and scope of the present disclosure. It will be apparent to those skilled in the relevant art that the present disclosure may also be used in various other applications.

[0022] It should be noted that the phrases "one embodiment," "an embodiment," "example embodiment," "some embodiments," "specific embodiments," etc., mentioned in the specification indicate that the described embodiments may include a particular feature, structure, or characteristic, but not every embodiment necessarily includes the particular feature, structure, or characteristic. In addition, these phrases do not necessarily refer to the same embodiment. Furthermore, when a particular feature, structure, or characteristic is described as being associated with an embodiment, whether or not explicitly described, those skilled in the relevant art may implement such feature, structure, or characteristic in conjunction with other embodiments.

[0023] In general, terms can be understood, at least in part, from their use in context. For example, as used herein, the term "one or more" can be used to describe any feature, structure, or characteristic in the singular, or can be used to describe a combination of features, structures, or characteristics in the plural, depending at least in part on the context. Similarly, depending at least in part on the context, the terms "a," "an," or "the" can also be understood to express singular usage or plural usage. Furthermore, depending at least in part on the context, the term "based on" can also be understood to not necessarily express an exclusive set of factors, but rather to allow for the presence of additional factors that are not necessarily explicitly described.

[0024] Various aspects of a wireless communication system will now be described with reference to various apparatuses and methods. These apparatuses and methods are described in the following detailed description and illustrated in the accompanying drawings as various blocks, modules, units, components, circuits, steps, operations, processes, algorithms, etc. (collectively, "elements"). These elements can be implemented using electronic hardware, firmware, computer software, or any combination thereof. Whether these elements are implemented as hardware, firmware, or software depends on the specific application and the design constraints imposed on the overall system.

[0025] The techniques described in this disclosure can be used in various wireless communication networks, such as code division multiple access (CDMA) systems, time division multiple access (TDMA) systems, frequency division multiple access (FDMA) systems, orthogonal frequency division multiple access (OFDMA) systems, single carrier frequency division multiple access (SC-FDMA) systems, wireless local area network (WLAN) systems, and other networks. The terms "network" and "system" are often used interchangeably. A CDMA network can implement a radio access technology (RAT) such as Universal Terrestrial Radio Access (UTRA), Evolved UTRA (E-UTRA), and CDMA 2000. A TDMA network can implement a RAT such as Global System for Mobile Communications (GSM). An OFDMA network can implement a RAT such as LTE or NR. A WLAN system can implement a RAT such as Wi-Fi. The techniques and systems described in this disclosure can be used in the wireless networks and RATs mentioned above, as well as other wireless networks and RATs.

[0026] Multiple-input, multiple-output (MIMO) technology forms the foundation of various wireless communication systems, such as 3G, LTE, and NR, to name a few. In a MIMO system, both the transmitter and receiver are equipped with multiple antennas. Multiple data streams can be transmitted simultaneously to the receiver through spatial multiplexing. Each data stream can be associated with a transport layer. Spatial multiplexing provides high spectral efficiency, but at the expense of increased signal processing complexity, particularly in the baseband chip's MIMO detector, which is used to recover the data streams from the channel and noise / interference.

[0027] For a MIMO system with N transmission layers and M receiving antennas, the mathematical system model can be described using formula (1):

[0028] y=Hx+n (1),

[0029] in, is the received signal vector, is the transmitted quadrature amplitude modulation (QAM) symbol vector, where is the set of possible QAM symbols for a particular modulation order m (e.g., for quadrature phase shift keying (QPSK), m=2; for 16QAM, m=4; for 64QAM, m=6; for 256QAM, m=8; for 1024QAM, m=10, etc.); is the complex channel matrix; is a Gaussian white noise vector.

[0030] After performing noise whitening on the received signal and the estimated channel, n can be a complex additive white Gaussian noise vector with variance unity and can be considered uncorrelated across vector elements, e.g., the noise covariance matrix Φ n =I M The baseband chip can be configured to perform MIMO detection to estimate the channel matrix given the whitening and the whitened received signal vector Estimate x in the case of .

[0031] Among various types of MIMO detection techniques, maximum likelihood detection (MLD) provides the best theoretical error performance. Using the above system model, assuming white noise Φ n =I N , the maximum likelihood (ML) solution is equivalent to solving the least squares problem of equation (2):

[0032]

[0033] in, is the estimated ML of the transmitted QAM symbol vector.

[0034] In a specific implementation, the least squares problem can be solved by using the QR decomposition technique, for example, by Perform QR decomposition, where is an orthogonal matrix and is an upper triangular matrix. It can be assumed, without loss of generality, that the number of receive antennas is greater than or equal to the number of transmit layers, i.e., M ≥ N. Under this assumption, the lower MN rows of the upper triangular matrix R are all zero, so only the upper N×N square matrix and the upper rows of the triangular matrix are meaningful. By using Q H Process the received vector y, for example, It is The sufficient statistics of , the system model can be transformed into the following equation (3):

[0035]

[0036] in Still a Therefore, the problem of finding the ML solution can be equivalent to solving equation (4):

[0037]

[0038] By way of example and not limitation, given a system with four transmission layers and four receiver antennas (e.g., a 4x4 system), equation (4) may be expressed as:

[0039]

[0040] For hard-output detection schemes, such as Vertical-Bell Laboratories Layered Space-Time (V-BLAST), detection starts from layer 3 by solving equation (5):

[0041]

[0042] Then, layer 2 can be detected by canceling the detection results of layer 3, as shown in the following equation (6):

[0043]

[0044] The continuous interference cancellation / inverse permutation process performed on layer 2 can be performed iteratively for layers 1 and 0. However, such techniques are susceptible to error propagation. That is, a detection error in layer k can negatively impact the detection of layers k-1, ..., 0. Therefore, the transmitted layers can be ordered so that the stronger layers are detected first to minimize the error propagation problem.

[0045] On the other hand, the soft output detection scheme based on ML detection is essentially The solution of equation (4) is searched in the vector space of . It can be shown that the problem in equation (4) can be transformed into an equivalent QAM symbol tree search problem, where the tree path from the root to the leaf represents a transmitted QAM symbol vector based at least in part on the upper triangular matrix R. Such a QAM symbol vector has a length of N and can include QAM symbols from those N transmission layers. The search result of one iteration of the tree search can be a path metric representing one or more QAM symbols.

[0046] The soft outputs (e.g., log-likelihood ratios (LLRs)) of the QAM symbol vectors can be based on the best path metric produced in the tree search (e.g., max-log-maximum-a-posteriori (max logMAP)), or a subset of the path metrics (e.g., approximate logMAP). The improvement in soft output system performance is achieved at the expense of higher computational complexity associated with the tree search. As the modulation order m and / or the number of transmission layers N increase, the size of the tree, and therefore the computational complexity of the ML detection algorithm, increases exponentially. Therefore, even for conventional near-MLD schemes that reduce computational complexity by limiting the number of paths in the tree search, the computational resources used to perform such calculations are still prohibitive for higher-order QAMs (e.g., 16QAM, 64QAM, 256QAM, 1024QAM, etc.).

[0047] There are different categories of MIMO detectors. These categories may include, for example, linear MIMO detectors (e.g., minimum mean squared error (MMSE) MIMO detectors, maximum ratio combining (MRC) MIMO detectors, zero forcing (ZF) MIMO detectors, etc.), sphere decoding-based MIMO detectors (e.g., breadth-first tree search MIMO detectors, depth-first tree search MIMO detectors, best-first tree search MIMO detectors, etc.), and interference cancellation-assisted MIMO detectors, to name a few. Among these categories, sphere decoding-based MIMO detectors offer design flexibility in the trade-off between near-optimal ML performance and reduced computational complexity.

[0048] In many implementations, the optimal solution for decoding spatially multiplexed signals is ML, which involves a multi-dimensional exhaustive tree search and is particularly beneficial for use in high-performance applications. Sphere decoding is an iterative method for computing ML estimates. Like ML algorithms, sphere decoding finds the grid point closest to the received vector, but the search is limited to a subset of constellation points that lies within a sphere centered on the received vector. Limiting the search to a subset of constellation points significantly reduces computational complexity. Therefore, sphere decoding offers a computationally efficient decoding algorithm with the performance of ML. However, one of the challenges in implementing sphere decoding is that the number of iterations per implementation is neither defined nor bounded. Consequently, traditional sphere decoding may not be suitable for certain hardware implementations.

[0049] One of the sphere decoding tree search algorithms is the traditional fixed complexity tree search algorithm. The fixed complexity tree search algorithm limits the total number of candidate constellation points tested in each layer N, thereby reducing the complexity of the calculations performed by the MIMO detector. The fixed complexity tree search MIMO detector provides a beneficial performance-complexity trade-off. An example of a traditional fixed complexity tree search algorithm is the so-called N-1-1-1 tree search. Figure 8 and Figure 9 Describe it.

[0050] Furthermore, in OFDMA systems, the reference resource elements (REs) used for channel and noise estimation are limited and dispersed in both time and frequency directions to limit system overhead. Furthermore, due to the nature of random interference, interference is typically unbalanced in both time and frequency directions. Therefore, averaging over frequency and time may not be a viable solution to improve estimation performance. Due to the limited number of available reference signals, conventional noise / interference covariance estimation techniques are far from accurate enough. One conventional technique also utilizes data REs in noise / interference covariance estimation. However, when performing noise covariance estimation, the data symbols are unknown, limiting the accuracy of the noise covariance estimation.

[0051] Figure 8 8. The lattice 800 is shown as a lattice that can be used in a conventional fixed complexity tree search. For example, the lattice 800 can be used for an N-1-1-1 tree search. Figure 8 As shown, the grid 800 includes an anchor point 802, a bottom layer N-1 804a, a second layer N-2 804b, a third layer N-3 804c, and a fourth layer 0 804d. Each of the layers 804a, 804b, 804c, 804d in the grid 800 includes a set of eight constellation points 810. Figure 8 In the example shown, the subset of selected constellation points 806 includes the middle six constellation points 806. The anchor point for the bottom layer N-1 804a is selected based on the QAM slice.

[0052] like Figure 8 As shown, the tree search begins with a bottom layer (e.g., layer N-1 804a) having a preselected subset of constellation points 806. Starting from layer N-1 804a, paths 808 are evolved by finding the best child node along each path. The child node is determined by direct constellation slicing after removing interference from previous layers. In the case of smaller modulation orders (e.g., QPSK, 16QAM, etc.), the N subsets of constellation points 806 selected in bottom layer N-1 804a can be the complete set of constellation points (not shown), or a subset in the case of larger modulation orders (e.g., 64QAM, 256QAM, 1024QAM, etc.).

[0053] The reduced number of candidate paths explored in N-1-1-1 or other reduced complexity tree search algorithms may result in insufficient statistics in the generated LLRs, which can degrade the performance of MIMO detectors, especially in correlated channels and / or high modulation orders. To improve LLR quality, multiple QR+ tree searches can be performed in parallel for each permutation of the transport layer ordering, thereby performing one QR+ tree search for each permutation. Figure 9 Describe any other details.

[0054] Figure 9A parallel architecture of a conventional MIMO detector 900 is shown that can perform multiple QR+tree searches in parallel. Figure 9 The conventional MIMO detector 900 shown may be part of a 4x4 system. Thus, four QR+ tree searches may be performed in parallel. Figure 9 As shown, the MIMO detector 900 may include a layer ordering component 902 configured to order different N transmission layers, thereby searching in parallel the four layer ordering combinations in which each transmission layer is searched. Each of the QR components 904a, 904b, 904c, 904d may be configured to perform a QR decomposition of the complex channel matrix H based on the corresponding layer ordering. Each of the QR components 904a, 904b, 904c, 904d outputs and (It is Q -1 y) can be input to their respective tree search units 906a, 906b, 906c, 906d, which can perform fixed complexity tree searches to determine path metrics based on the layer ordering. Each of the four selected path metrics can be input to an LLR component 908, which is configured to estimate the QAM symbol(s) of the received data stream. However, statistical errors associated with path metrics for transmission layers (e.g., layer 0, layer 1, etc.) with weaker signal power can limit the accuracy of the LLRs generated for the QAM symbols, thereby limiting the performance of baseband chips with such MIMO detectors.

[0055] Therefore, there is an unmet need for a baseband chip that can perform multiple QR+tree searches with reduced complexity and improved LLR accuracy compared to the above-mentioned conventional techniques.

[0056] The baseband chip of the present disclosure provides a solution by sequentially performing multiple QR+ tree searches. That is, the baseband chip described below can be configured to perform recursive QR+ tree searches. The tree searches in the later stages associated with the low-power transmission layers (e.g., layer 0, layer 1, layer 2, etc.) can utilize the statistics generated in the previous (multiple) tree searches for the high-power transmission layers (e.g., layer 3, layer 2, layer 1), thereby improving the accuracy of the LLR calculation, for example, as described below in conjunction with Figure 3 Description.

[0057] Given the recursive QR+tree search framework of the present disclosure, successive iterations can leverage previous iterations by using estimated data symbols to help improve noise / interference estimates, for example, as described below in conjunction with Figure 4 Description.

[0058] Figure 1 1 shows an exemplary wireless network 100 in which certain aspects of the present disclosure may be implemented according to some embodiments of the present disclosure. Figure 1 As shown, wireless network 100 may include a network of nodes, such as user equipment (UE) 102, access node 104, and core network element 106. UE 102 may be any terminal device, such as a mobile phone, desktop computer, laptop computer, tablet computer, in-vehicle computer, game console, printer, positioning device, wearable electronic device, smart sensor, or any other device capable of receiving, processing, and transmitting information, such as any member of a vehicle-to-everything (V2X) network, a cluster network, a smart grid node, or an Internet of Things (IoT) node. It will be understood that UE 102 is shown as a mobile phone for illustration purposes only and is not intended to be limiting.

[0059] The access node 104 may be a device that communicates with the user equipment 102, such as a wireless access point, a base station (BS), a node B (NodeB), an enhanced base station (eNodeB or eNB), a next-generation base station (gNodeB or gNB), a cluster master node, etc. The access node 104 may have a wired connection to the user equipment 102, a wireless connection to the user equipment 102, or any combination thereof. The access node 104 may be connected to the user equipment 102 through multiple connections, and the user equipment 102 may be connected to other access nodes in addition to the access node 104. The access node 104 may also be connected to other user equipment. It will be understood that the access node 104 is shown as a wireless tower for illustration and not as a limitation.

[0060] The core network element 106 can serve the node 104 and the user equipment 102 to provide core network services. Examples of the core network element 106 may include a home subscriber server (HSS), a mobility management entity (MME), a serving gateway (SGW), or a packet data network gateway (PGW). These are examples of core network elements of an evolved packet core (EPC) system, which is the core network of an LTE system. Other core network elements can be used in LTE and other communication systems. In some embodiments, the core network element 106 includes an access and mobility management function (AMF) device, a session management function (SMF) device, or a user plane function (UPF) device of the core network of the NR system. It will be understood that the core network element 106 is shown as a group of rack-mounted servers for illustration and not limitation.

[0061] Core network element 106 can be connected to a larger network, such as the Internet 108 or another Internet Protocol (IP) network, to transmit packet data over any distance. In this manner, data from user device 102 can be transmitted to other user devices connected to other access points, including, for example, computer 110 connected to Internet 108 using a wired or wireless connection, or tablet computer 112 connected wirelessly to Internet 108 via router 114. Thus, computer 110 and tablet computer 112 provide additional examples of possible user devices, and router 114 provides another example of a possible access node.

[0062] A general example of a rack-mounted server is provided as an illustration of the core network element 106. However, there may be multiple elements in the core network, including database servers, such as database 116, and security and authentication servers, such as authentication server 118. For example, database 116 may manage data related to user subscriptions to network services. A home location register (HLR) is an example of a standardized database of user information for a cellular network. Similarly, authentication server 118 may handle authentication of users, sessions, etc. In an NR system, an authentication server function (AUSF) device may be a specific entity that performs authentication of user equipment. In some embodiments, a single server rack may handle multiple such functions so that the connections between the core network element 106, authentication server 118, and database 116 may be local connections within a single rack.

[0063] Figure 1 Each element of can be considered as a node of wireless network 100. More details on possible implementations of nodes are provided below. Figure 7 The description of the node 700 is provided as an example. The node 700 can be configured as Figure 1 Similarly, the node 700 may also be configured as a user equipment 102, an access node 104 or a core network element 106. Figure 1 Computer 110, router 114, tablet computer 112, database 116 or authentication server 118 in Figure 7 As shown, node 700 may include a processor 702, a memory 704, and a transceiver 706. These components are shown as being connected to each other via a bus, but other connection types are also possible. When node 700 is a user device 102, it may also include additional components, such as a user interface (UI), sensors, etc. Similarly, when node 700 is configured as a core network element 106, node 700 may be implemented as a blade in a server system. Other implementations are also possible.

[0064] The transceiver 706 may include any suitable device for sending and / or receiving data. Although only one transceiver 706 is shown for simplicity of illustration, the node 700 may include one or more transceivers. Antenna 708 is shown as a possible communication mechanism for the node 700. Multiple antennas and / or antenna arrays may be used to receive multiple spatially multiplexed data streams. In addition, examples of the node 700 may use wired technology instead of (or in addition to) wireless technology to communicate. For example, the access node 104 may communicate wirelessly with the user device 102 and may communicate with the core network element 106 via a wired connection (e.g., via an optical cable or a coaxial cable). Other communication hardware, such as a network interface card (NIC), may also be included.

[0065] like Figure 7 As shown, node 700 may include a processor 702. Although only one processor is shown, it is understood that multiple processors may be included. Processor 702 may include a microprocessor, a microcontroller unit (MCU), a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), a programmable logic device (PLD), a state machine, gating logic, discrete hardware circuits, and other suitable hardware configured to perform the various functions described in this disclosure. Processor 702 may be a hardware device with one or more processing cores. Processor 702 may execute software. Whether referred to as software, firmware, middleware, microcode, hardware description language, or otherwise, software should be broadly interpreted to mean instructions, instruction sets, code, code segments, program code, programs, subroutines, software modules, applications, software applications, software packages, routines, subroutines, objects, executable files, execution threads, processes, functions, and the like. Software may include computer instructions written in interpreted languages, compiled languages, or machine code. Under the broad category of software, other techniques for indicating hardware are also permitted.

[0066] like Figure 7As shown, node 700 may also include a memory 704. Although only one memory is shown, it is understood that multiple memories may be included. Memory 704 can broadly include both internal memory and storage. For example, memory 704 may include random access memory (RAM), read-only memory (ROM), static RAM (SRAM), dynamic RAM (DRAM), ferroelectric RAM (FRAM), electrically erasable programmable ROM (EEPROM), CD-ROM or other optical disk storage, a hard disk drive (HDD), such as a magnetic disk storage or other magnetic storage device, a flash drive, a solid-state drive (SSD), or any other medium that can be used to carry or store the desired program code in the form of instructions that can be accessed and executed by processor 702. Broadly speaking, memory 704 can be implemented by any computer-readable medium, such as a non-transitory computer-readable medium.

[0067] The processor 702, memory 704, and transceiver 706 can be implemented in various forms in the node 700 to perform wireless communication functions. In some embodiments, the processor 702, memory 704, and transceiver 706 of the node 700 are implemented (e.g., by integration) on one or more system-on-chips (SoCs). In one example, the processor 702 and memory 704 can be integrated into an application processor (AP) SoC (sometimes referred to as a "host," herein referred to as a "host chip") that is responsible for application processing in an operating system (OS) environment that generates raw data to be transmitted. In another example, the processor 702 and memory 704 can be integrated into a baseband processor (BP) SoC (sometimes referred to as a "modem," herein referred to as a "baseband chip") that is used to convert raw data from, for example, a host into a signal that can be used to modulate a transmission carrier frequency, and vice versa, so that a real-time operating system (RTOS) can be run. In yet another example, the processor 702 and the transceiver 706 (and in some cases, the memory 704) can be integrated into an RF SoC (sometimes referred to as a "transceiver," herein referred to as an "RF chip") that transmits and receives RF signals. It will be appreciated that in some examples, some or all of the host chip, baseband chip, and RF chip can be integrated into a single SoC. For example, the baseband chip and the RF chip can be integrated into a single SoC that manages all radio functions for cellular communications.

[0068] Return Reference Figure 1As described in detail below, in some embodiments, any suitable node of wireless network 100 (e.g., user device 102 or access node 104) can process a MIMO data stream by performing a recursive QR+ tree search when transmitting a signal to another node, e.g., from user device 102 through access node 104, or vice versa. Thus, the present application can improve the accuracy of LLR calculation while reducing computational complexity compared to known solutions that do not use statistics generated by transmission layers received at higher power when subsequently performing a tree search on transmission layers received at lower power.

[0069] Figure 2 FIG2 shows a block diagram of an apparatus 200 including a baseband chip 202, an RF chip 204, and a host chip 206 according to some embodiments of the present disclosure. The apparatus 200 may be Figure 1 Examples of any suitable node in wireless network 100, such as user equipment 102 or access node 104. Figure 2 As shown, the device 200 may include a baseband chip 202, an RF chip 204, a host chip 206, and one or more antennas 210. In some embodiments, as described below with respect to Figure 7 As described, the baseband chip 202 is implemented by a processor 702 and a memory 704, and the RF chip 204 is implemented by a processor 702, a memory 704, and a transceiver 706. In addition to the on-chip memory (also referred to as "internal memory", such as registers, buffers, or caches) on each chip 202, 204, or 206, the device 200 may also include an external memory 208 (e.g., system memory or main memory), which may be shared by each chip 202, 204, or 206 via a system / main bus. Although in Figure 2 The baseband chip 202 is shown as a standalone SoC, but it can be understood that in one example, the baseband chip 202 and the RF chip 204 can be integrated into one SoC; in another example, the baseband chip 202 and the host chip 206 can be integrated into one SoC; in yet another example, the baseband chip 202, the RF chip 204 and the host chip 206 can be integrated into one SoC.

[0070] In the uplink, the host chip 206 can generate raw data and send it to the baseband chip 202 for encoding, modulation, and mapping. The baseband chip 202 can also access the raw data generated by the host chip 206 and stored in the external memory 208, for example, using direct memory access (DMA). The baseband chip 202 can first encode (for example, through source coding and / or channel coding) the raw data and modulate the coded data using any suitable modulation technology, such as multi-phase pre-shared key (MPSK) modulation or orthogonal amplitude modulation (QAM). The baseband chip 202 can perform any other functions, such as symbol or layer mapping, to convert the raw data into a signal that can be used to modulate the carrier frequency for transmission. In the uplink, the baseband chip 202 can send the modulated signal to the RF chip 204. The RF chip 204 can convert the modulated signal in digital form into an analog signal, i.e., an RF signal, through a transmitter (Tx) and perform any suitable front-end RF functions, such as filtering, up-conversion, or sampling rate conversion. The antenna 210 (eg, an antenna array) may transmit RF signals provided by a transmitter of the RF chip 204 .

[0071] In the downlink, the antenna 210 can receive the RF signal and pass the RF signal to the receiver (Rx) of the RF chip 204. The RF chip 204 can perform any suitable front-end RF functions, such as filtering, down-conversion, or sample rate conversion, and convert the RF signal into a low-frequency digital signal (baseband signal) that can be processed by the baseband chip 202. In the downlink, the baseband chip 202 can demodulate and decode the baseband signal to extract the original data that can be processed by the host chip 206. The baseband chip 202 can perform additional functions, such as error checking, demapping, channel estimation, descrambling, etc. The original data provided by the baseband chip 202 can be sent directly to the host chip 206 or stored in the external memory 208.

[0072] In some embodiments, the baseband chip 202 can perform operations related to the recursive QR+ tree search described below, for example, Figure 3-7 These operations may be performed by one or more functional blocks of the baseband chip 202. For example, although Figure 2Although not shown, the baseband chip 202 may include, for example, one or more layer sorting components configured to sort transmission layers according to received signal strength; one or more QR decomposition units configured to perform QR decomposition of the channel matrix; one or more subset selection units configured to select a subset of constellation points to search; one or more tree search units configured to determine a path metric based at least in part on the subset of constellation points; one or more anchor point selection units configured to select an anchor point for a subsequent tree search for another transmission layer, and / or an LLR unit configured to determine an LLR based on a path metric determined from each tree search. Furthermore, the anchor point selection unit of one iteration may input information associated with the selected anchor point into the subset selection unit associated with the subsequent iteration.

[0073] Therefore, compared to the known solution that does not use the statistical data generated by the transmission layer received at higher power in the subsequent tree search, the present application can improve the accuracy of LLR calculation while reducing the computational complexity. Figure 3-7 Further details of the recursive QR+ tree search of the present disclosure are described.

[0074] Figure 3 A block diagram of a first exemplary baseband chip 300 according to some embodiments of the present disclosure is shown. The baseband chip 300 may be a node equipped with multiple antennas and configured for MIMO wireless communication, such as the UE 102 or the access node 104. In an example embodiment, the functional units described below may be part of a MIMO detector and / or a MIMO demapper that is part of the baseband chip 300. Although the baseband chip 300 will be described in conjunction with a 4x4 MIMO system, the recursive QR+ tree search described below may be applicable to systems with different numbers of receive antennas and / or data streams, depending on the number of bits transmitted per symbol, without departing from the scope of the present disclosure.

[0075] refer to Figure 3 The baseband chip 300 may include various functional units configured to perform various iterations of the recursive QR+ tree search, so that the path metrics and / or detected paths of one iteration can be used for anchor point and subset selection of the bottom layer of a subsequent iteration. In the first iteration, the transmission layers may be sorted from top to bottom in ascending order of power.

[0076] For example, layer 0 can be received with the lowest signal power, layer 1 can be received with the next highest signal power, layer 2 can be received with the next highest signal power, and layer 3 can be received with the highest signal power. In a first iteration, the first order from top to bottom can be, for example, layer 0, layer 1, layer 2, layer 3. In a second iteration, the second order from top to bottom can be, for example, layer 0, layer 1, layer 3, layer 2. In a third iteration, the third order from top to bottom can be, for example, layer 0, layer 2, layer 3, layer 1. In a fourth iteration, the fourth order from top to bottom can be, for example, layer 1, layer 2, layer 3, layer 0.

[0077] In a specific implementation, each iteration may include a set of dedicated functional blocks. The functional blocks may include, but are not limited to, layer ordering units 302a, 302b, 302c, 302d, QR decomposition units 304a, 304b, 304c, 304d, subset selection units 306a, 306b, 306c, 306d, tree search units 308a, 308b, 308c, 308d, and path metric units 310a, 310b, 310c. The last iteration may include an LLR unit 310d configured to generate LLRs associated with the estimated QAM symbols.

[0078] First, the baseband chip 300 can perform QAM slicing to select an anchor point from the constellation associated with layer 3. Since layer 3 has the highest signal strength, using QAM slices of layer 3 to select an anchor point will result in less ambiguity compared to QAM slices of layers 2, 1, and 0.

[0079] The recursive QR+ tree search may start from the first iteration. Here, the first layer sorting unit 302a may sort each transmission layer in ascending order of signal strength. The signal strength of the transmission layer may be determined based at least in part on a signal characteristic corresponding to the received signal power. By way of example and not limitation, the signal characteristic may include one or more of a signal-to-noise ratio (SNR), a signal-to-interference-plus-noise ratio (SINR), a received signal strength indication (RSSI), to name a few. In a particular embodiment, the first layer sorting unit 302a may determine the signal strength of each transmission layer. However, in other particular embodiments, the signal strength may be determined by Figure 3 In such an embodiment, information associated with the signal strength of each transmission layer may be input into the first layer sorting unit 302a. In either implementation, the first layer sorting unit 302a may generate a first complex channel matrix based at least in part on the first sorting of the transmission layers.

[0080] The first complex channel matrix H1 may be input to the first QR decomposition unit 304a. The first QR decomposition unit 304a may perform QR decomposition of the first complex channel matrix H1 to generate a first estimated channel matrix It can be input into the first subset selection unit 306a. Can also be generated as a by-product of QR decomposition.

[0081] The first subset selection unit 306a may select a subset of constellation points around a layer 3 anchor point. The subset of constellation points may be selected based on predetermined criteria. In other specific embodiments, the subset of constellation points may be selected based on a box around the anchor point. In another embodiment, the subset of constellation points may be selected based on a distance from the anchor point. The size of the subset may be determined based on, for example, SINR, RSSI, etc.

[0082] It can be connected to the layer 3 anchor point, the constellation point subset in layer 3, and receiving signals The relevant information is input to the first tree search unit 308a. The first tree search unit 308a can generate a first search tree based on the first ranking of the transport layer. The first tree search unit 308a can use the first search tree to perform a tree search to determine various path metrics.

[0083] For example, a search tree (also known as a "decoding tree" or "logic tree") is a lattice structure representing constellation points of a received signal, corresponding to, for example, a 16QAM constellation. The lattice structure includes a plurality of nodes, each of which includes an integer value representing a symbol component of a received data signal. In certain embodiments, the received data signal can be represented according to a real-valued or complex-valued representation. As used herein, the term "node" is used to designate a symbol component of a received data signal.

[0084] The first node of the search tree is called the "root node". Nodes that do not have any child nodes (child nodes are also called "successors") are called "leaf" nodes and correspond to the lowest level in the search tree. The root node is the highest node in the search tree and has no parent node. The depth (or dimension) of a given node represents the length of the path from that given node to the root node in the search tree. All nodes of the search tree can be reached from the root node. Therefore, each path from the root node to a leaf node represents a possible transmission signal. The nodes in the search tree represent symbols s i Different possible values ​​of , where s i The i in the string represents an integer from n to 1, and s i represents the real and imaginary parts of the transmitted information vector. A leaf node specifies a node with a depth of n. According to the notation used here, a node s k The child nodes are composed of components s k-1Specifies that the path at depth i in the search tree is represented by a vector s of length (n-i+1) (i) =(s n , s n-1 ,...s i ) specified.

[0085] The sequential decoding algorithm is a tree search based decoding algorithm based on a tree representation of an ML optimization problem (represented by a search tree). This sequential decoding algorithm uses a stack or list to store the best candidate nodes. The sequential decoding technique considers the path metric of each expanded node of the search tree. Each node selected in the expanded node (component of the candidate grid) is stored in the stack in association with the calculated path metric. The path metric is typically determined as a function of the Euclidean distance between the received signal and the symbol vector represented by the path metric between the root node and the current node. However, different functions may be used to determine the path metric without departing from the scope of the present disclosure.

[0086] The various path metrics determined by the first tree search unit 308a may be input to the first path metric unit 310a. The first path metric unit 310a may select the path metric having the smallest Euclidean distance as the path metric input to the LLR unit 310d for layer 3.

[0087] The first path metric unit 310a can select a layer 2 anchor point for use in the second iteration. In some embodiments, the layer 2 anchor point can be selected based at least in part on, for example, various path metrics and / or the path input by the first tree search unit 308a. For example, a layer 2 node in the best path can be selected as the anchor point. Information associated with the layer 2 anchor point can be input into the second subset selection unit 306b. Because the signal power associated with layer 3 is higher than that associated with layer 2, using the path metric from the first iteration and / or selecting the layer 2 anchor point can improve the accuracy of the path metric determined in the second iteration.

[0088] In the second iteration, the second layer sorting unit 302b may generate a second complex channel matrix H2 based on a second sorting in which layer 3 and layer 2 are transposed, compared to the sorting in the first iteration. That is, in the second iteration, layer 2 is the lowest layer, and layer 3 is the next lowest layer. In certain embodiments, the second layer sorting unit 302b may determine the signal strength of each transmission layer. In other certain embodiments, the first layer sorting unit 302a may input information associated with the signal power of each transmission layer into the other layer sorting units 302b, 302c, and 302d.

[0089] Regarding the second iteration, the second layer ordering unit 302b may generate a second complex channel matrix based at least in part on the second ordering of the transmission layers:

[0090] The second complex channel matrix H2 can be input to the second QR decomposition unit 304b. The second QR decomposition unit 304b can perform QR decomposition on the second complex channel matrix H2 to generate a second estimated channel matrix and receiving signals This may be input to the second subset selection unit 306b.

[0091] The second subset selection unit 306b may select a subset of constellation points around the anchor point of layer 2. The selection of the subset of constellation points may use the same or similar operations as those described above in connection with the selection of the anchor point of layer 3.

[0092] The information associated with the layer 2 anchor point, the subset of constellation points selected for layer 2, and The second tree search unit 308b may generate a second search tree based on the second ranking of the transport layers and may use the second search tree to perform a tree search to determine various path metrics associated with the layer 2 anchor point.

[0093] The second iteration may conclude with the second path metric unit 310b selecting the best path metric and associated path based on the second ranking. The second ranked path metrics may be input to the LLR unit 310d. In addition, the second path metric unit 310b may select a layer 1 anchor point. Information associated with the layer 1 anchor point may be input to the third subset selection unit 306c for selecting a subset of constellation points based on the third ranking.

[0094] The various path metrics determined by the second tree search unit 308b may be input to the second path metric unit 310b. The second path metric unit 310b may select the path metric having the smallest Euclidean distance as the path metric input to the LLR unit 310d for layer 2.

[0095] The second path metric unit 310b may select a layer 1 anchor point for use in the third iteration. In certain embodiments, the layer 1 anchor point may be selected based at least in part on various path metrics and paths, for example, input by the second tree search unit 308b. For example, in certain embodiments, a layer 1 node in the best path may be selected as the anchor point. In another embodiment, the layer 1 anchor point may be selected based at least in part on various path metrics, for example, input by the first tree search unit 308a and the second tree search unit 308b. Here, a layer 1 node in the best path from the first two iterations may be selected as the anchor point. Information associated with the layer 1 anchor point may be input into the third subset selection unit 306c. Because the signal power associated with layer 2 is higher than that of layer 1, selecting the layer 1 anchor point using the path metric from the second iteration may improve the accuracy of the path metric determined in the third iteration.

[0096] Each functional unit associated with the third and fourth iterations may perform operations identical or similar to those described above in conjunction with the first and second iterations. Therefore, for the sake of brevity, the description of these operations performed by the functional blocks associated with the third and fourth iterations will not be repeated.

[0097] After performing each of the four iterations, the LLR unit 310d may generate LLRs associated with one or more QAM symbols associated with each layer. Compared to known schemes that do not use statistics generated by transmission layers received at higher power when subsequently performing tree searches on transmission layers received at lower power, using the combined Figure 3 The described techniques may increase the accuracy of the LLR calculation(s) performed by LLR unit 310d while still providing reduced computational complexity.

[0098] Figure 4 A block diagram of a second exemplary baseband chip 400 according to some embodiments of the present disclosure is shown. The baseband chip 400 may be a node equipped with multiple antennas and configured for MIMO wireless communication, such as the UE 102 or the access node 104. In an exemplary embodiment, the functional units described below may be part of a MIMO detector and / or a MIMO demapper that is part of the baseband chip 400. Although the baseband chip 400 will be described in conjunction with a 4x4 MIMO system, the recursive QR+ tree search described below may be applicable to systems with different numbers of receive antennas and / or data streams, depending on the number of bits transmitted per symbol, without departing from the scope of the present disclosure.

[0099] refer to Figure 4The baseband chip 400 may include various functional units configured to perform various iterations of the recursive QR+ tree search, so that the underlying path metrics of one iteration can be used for anchor point and subset selection of the underlying layers of subsequent iterations. In the first iteration, the transport layers may be sorted from top to bottom in ascending order of power.

[0100] For example, layer 0 can be received with the lowest signal power, layer 1 can be received with the next highest signal power, layer 2 can be received with the next highest signal power, and layer 3 can be received with the highest signal power. In a first iteration, the first order from top to bottom can be, for example, layer 0, layer 1, layer 2, layer 3. In a second iteration, the second order from top to bottom can be, for example, layer 0, layer 1, layer 3, layer 2. In a third iteration, the third order from top to bottom can be, for example, layer 0, layer 2, layer 3, layer 1. In a fourth iteration, the fourth order from top to bottom can be, for example, layer 1, layer 2, layer 3, layer 0.

[0101] In a specific implementation, each iteration may include a set of dedicated functional blocks. The functional blocks may include, but are not limited to, noise estimation units 402a, 402b, 402c, 402d, noise whitening units 404a, 404b, 404c, 404d, QR decomposition units 406a, 406b, 406c, 406d, tree search units 408a, 408b, 408c, 408d, and path metric units 410a, 410b, 410c. The last iteration may include an LLR unit 410d, which is configured to generate LLRs associated with the estimated data RE. Figure 4 In the above, you can use Figure 3 The same or similar techniques are described to determine layer ordering.

[0102] Figure 4 The recursive QR+ tree search in [ 0 ] can start from the first iteration. Here, the first noise estimation unit 402a can perform the first noise estimation using the reference REs in layer 3. Without departing from the scope of the present disclosure, any technique known in the art can be used to perform noise estimation based on reference REs. Information associated with the first noise estimation can be input into the first noise whitening unit 404a.

[0103] The first noise whitening unit 404a can perform noise whitening on the channel matrix of the first ranking and received signal based at least in part on the first noise estimate. A whitening transform is a linear transformation that transforms a noise vector with a known covariance matrix into a set of new variables with a covariance of the identity matrix. That is, the new variables are uncorrelated and each has a variance of 1. Because this transformation transforms the input noise into a white noise vector, it is referred to as "whitening."

[0104] For example, suppose X is a random noise vector with non-singular covariance matrix Σ and mean 0. Then, use the condition W H W=Σ -1 The transformation Y=WX of the whitening matrix W produces a whitened random vector Y with unit diagonal covariance.

[0105] The output of the first noise whitening unit 404a may be a first complex channel matrix H1 associated with the first ordering of the transmission layer and a whitened received signal Y. The first complex channel matrix H1 and Y may be input to a first QR decomposition unit 406a. The first QR decomposition unit 406a may perform QR decomposition on the first complex channel matrix H1 to generate a first estimated channel matrix and receive signals This may be input to the first tree search unit 408a.

[0106] The first tree search unit 408a may be based on the first estimated channel matrix and receive signals Generate a first search tree. The first tree search unit 408a can use the first search tree to perform a tree search to determine various paths associated with path metrics, which can be input into the first path metric unit 410a. The first path metric unit 410a can select the best path with the least path metric. The best path selected by the first path metric unit 410a can be associated with the estimated data RE in the resource block (RB) (also known as "data block"). The first path metric unit 410a can use any hard decision technique known in the art to estimate the data symbols in each layer. The estimated data RE is then reconstructed based on the estimated data symbols and the estimated channel in each layer. Information associated with the path metric selected by the first path metric unit 410a can be input into the LLR unit 410d.

[0107] The above-described technique in conjunction with the first iteration can be performed for each data RE in the data block received in layer 3. The first path metric unit 410a can then input information associated with each estimated data RE into the second noise estimation unit 402b. In a spatial multiplexing scenario, each RE can include four layers of data. More specifically, in the tree search graph, each path contains four nodes, and each node can correspond to a symbol (e.g., a constellation point) in a single layer.

[0108] In the second iteration, the second noise estimation unit 402b can perform channel noise estimation based at least in part on the reference REs and the estimated data REs input by the first path metric unit 410a. By using the estimated data REs and the sparse reference REs, the noise estimated by the second noise estimation unit 402b can have improved accuracy compared to conventional systems that rely solely on reference REs for noise estimation. Information associated with the second noise estimate can be input to the second noise whitening unit 404b.

[0109] The second noise whitening unit 404b may perform noise whitening of the second ordered channel matrix based at least in part on the second noise estimate and the received signal.Additional details of noise whitening are provided above in conjunction with the first noise whitening unit 404a.

[0110] The output of the second noise whitening unit 404b may be a second complex channel matrix H2 associated with the second ordered and whitened received signal Y of the transmission layer. The second complex channel matrix H2 and Y may be input to a second QR decomposition unit 406b. The second QR decomposition unit 406b may perform QR decomposition of the second complex channel matrix H2 to generate a second estimated channel matrix and receiving signals This may be input to the second tree search unit 408b.

[0111] The second tree search unit 408b may be based on the second estimated channel matrix A second search tree is generated. Second tree search unit 408b can perform a tree search using the second search tree to determine various paths and associated path metrics that can be input into second path metric unit 410b. Second path metric unit 410b can select the best path with the best path metric from the detected paths from tree search blocks 408a and 408b. The path selected by second path metric unit 410b can be associated with estimated data REs in the RB. In other words, the estimated data REs can be reconstructed from data symbols in each layer based on the best path and the estimated channel. Information associated with the path selected by second path metric unit 410b can be input into LLR unit 410d.

[0112] The techniques described above in connection with the second iteration may be performed on each data RE in the data block received in layer 2. The second path metric unit 410b may then input information associated with each estimated data RE into the third noise estimation unit 402c.

[0113] Each of the functional units associated with the third and fourth iterations may perform operations identical or similar to those described above in conjunction with the first and second iterations. Therefore, for the sake of brevity, the description of these operations performed by the functional blocks associated with the third and fourth iterations will not be repeated.

[0114] After performing each of the four iterations, the LLR unit 410d can generate LLRs for the data REs. Compared to conventional techniques that use only sparse reference REs for noise estimation, using the combined Figure 4 In successive iterations described, the baseband chip 400 is able to improve the noise / interference estimate using the estimation data RE from the previous iteration.

[0115] Figure 5 Flowchart showing a first exemplary method 500 for wireless communication according to some embodiments of the present disclosure. Examples of apparatuses that may perform the operations of method 500 include, for example Figure 3 The baseband chip 300 depicted in the figure or any other suitable device disclosed herein. It is understood that the operations shown in the method 500 are not exhaustive, and other operations may be performed before, after, or between any of the illustrated operations. In addition, some operations may be performed simultaneously or in different order. Figure 5 Execute in the order shown.

[0116] refer to Figure 5 At 502, a baseband chip may receive a data stream associated with a channel. In certain aspects, the channel may include multiple transmission layers (e.g., layer 3, layer 2, layer 1, and layer 0). In other certain aspects, the multiple transmission layers may include a first transmission layer (e.g., layer 3) associated with a first constellation point and a second transmission layer (e.g., layer 2) associated with a second constellation point. For example, referring to Figure 3 The baseband chip 300 may be configured for MIMO wireless communication and may receive bit streams in multiple transmission layers, such as layer 3, layer 2, layer 1, layer 0, etc. Each bit stream may include one or more QAM symbols.

[0117] At 504, the baseband chip may select a first anchor point from the first constellation point of the first transmission layer. Figure 3 , the baseband chip 300 can perform QAM slicing to select an anchor point from the constellation associated with layer 3. Since layer 3 has the highest signal strength, using QAM slices of layer 3 to select an anchor point will produce less ambiguity compared to QAM slices of layers 2, 1, and 0.

[0118] At 506, the baseband chip may select a first subset of constellation points from the first constellation point and the second constellation point based at least in part on the first anchor point. Figure 3, the first subset selection unit 306a may select a subset of constellation points around the layer 3 anchor point. The subset of constellation points may be selected based on a predetermined criterion. In other specific embodiments, the subset of constellation points may be selected based on a box around the anchor point. In another embodiment, the subset of constellation points may be selected based on the distance from the anchor point. The size of the subset may be determined based on, for example, SINR, RSSI, etc.

[0119] At 508, the baseband chip may perform a first iteration of a recursive tree search operation based at least in part on a first subset of constellation points associated with the first anchor point. Figure 3 , the recursive QR+ tree search may begin from the first iteration. Here, the first layer sorting unit 302a may sort each transmission layer in ascending order of signal strength. The signal strength of the transmission layer may be determined based at least in part on a signal characteristic corresponding to the received signal power. The first layer sorting unit 302a may generate a first complex channel matrix based at least in part on the first sorting of the transmission layers. The first complex channel matrix H1 may be input into the first QR decomposition unit 304a. The first QR decomposition unit 304a may perform QR decomposition of the first complex channel matrix H1 to generate a first estimated channel matrix and receiving signals This can be input to the first subset selection unit 306a. The first subset selection unit 306a can select a subset of constellation points around the layer 3 anchor point. The subset of constellation points can be selected based on a predetermined criterion. In other specific embodiments, the subset of constellation points can be selected based on a box around the anchor point. In another embodiment, the subset of constellation points can be selected based on the distance from the anchor point. and receiving signals The relevant information is input to the first tree search unit 308a. The first tree search unit 308a can generate a first search tree based on the first ranking of the transport layer. The first tree search unit 308a can use the first search tree to perform a tree search to determine various paths and associated path metrics.

[0120] At 510, the baseband chip may determine a first path based at least in part on a first iteration of a recursive tree search operation. Figure 3 The various paths and associated path metrics determined by the first tree search unit 308a may be input into the first path metric unit 310a. The first path metric unit 310a may select the best path based on the path metrics.

[0121] At 512, the baseband chip may select a second anchor point from the second constellation point of the second transmission layer based at least in part on the first path metric. In certain aspects, the second anchor point may be associated with a second iteration of the recursive tree search operation. For example, referring to Figure 3 , the first path metric unit 310a may select a layer 2 anchor point for use in the second iteration. In particular embodiments, the layer 2 anchor point may be selected based at least in part on various path metrics, for example, input by the first tree search unit 308a. For example, a layer 2 node in the best path may be selected as the anchor point in the first iteration. Information associated with the layer 2 anchor point may be input into the second subset selection unit 306b. Because the signal power associated with layer 3 is higher than that of layer 2, selecting the layer 2 anchor point using the path metric from the first iteration may improve the accuracy of the path metric determined in the second iteration.

[0122] At 514, the baseband chip may select a second subset of constellation points from the second constellation for layer 2 based at least in part on the second anchor point. Figure 3 , the second subset selection unit 306b may select a subset of constellation points around the anchor point of layer 2. The selection of the constellation point subset may use the same or similar operations as those described above in conjunction with the selection of the anchor point of layer 3.

[0123] At 516, the baseband chip may perform a second iteration of the recursive tree search operation based at least in part on the second subset of constellation points. Figure 3 , information associated with the subset of constellation points selected for layer 2, and receive signals The second tree search unit 308b may be input to the second tree search unit 308b. The second tree search unit 308b may generate a second search tree based on the second ranking of the transport layers. The second tree search unit 308b may use the second search tree to perform a tree search to determine various paths and associated path metrics.

[0124] At 518, the baseband chip may determine a second path metric based at least in part on the second iteration of the recursive tree search operation. Figure 3 The second iteration may end with the second path metric unit 310b selecting the best path metric based on the second ranking. The second ranked path metrics may be input to the LLR unit 310d.

[0125] At 520, the baseband chip may generate LLRs associated with the data stream based at least in part on the first path metric and the second path metric. Figure 3 After performing each of the four iterations, LLR unit 310d may generate LLRs associated with one or more QAM symbols associated with each layer.

[0126] Figure 6 6. A flow chart of a second exemplary method 600 for wireless communication according to some embodiments of the present disclosure is shown. Examples of apparatuses that may perform the operations of method 600 include, for example Figure 4 The baseband chip 400 depicted in the figure or any other suitable device disclosed herein. It is understood that the operations shown in method 600 are not exhaustive and other operations may be performed before, after or between any of the illustrated operations. In addition, some operations may be performed simultaneously or in different order. Figure 6 Execute in the order shown.

[0127] refer to Figure 6 At 602, a baseband chip may receive a data stream associated with a channel. In certain aspects, the channel may include a plurality of transmission layers. In other certain aspects, the plurality of transmission layers may include a first transmission layer and a second transmission layer associated with a resource block (RB). In other certain aspects, the RB may include at least one first reference RE and a plurality of first data resource elements (REs). For example, a reference Figure 4 , the baseband chip 400 can receive data streams including different resource blocks with different REs received through different transmission layers in a channel.

[0128] At 604, the baseband chip may perform an initial noise estimation of the channel based at least in part on the reference REs. Figure 4 , the first noise estimation unit 402a can perform a first noise estimation using the reference REs in layer 3. Without departing from the scope of the present disclosure, any technique known in the art can be used to perform noise estimation based on reference REs. Information associated with the first noise estimation can be input into the first noise whitening unit 404a.

[0129] At 606, the baseband chip may perform a first noise whitening operation on the channel and the received signal based at least in part on the initial noise estimate. Figure 4 , the first noise whitening unit 404a can perform noise whitening of the first sorted channel matrix based at least in part on the first noise estimate. A whitening transform is a linear transform that transforms a noise vector with a known covariance matrix into a set of new variables with a covariance of the unit matrix. That is, this set of new variables is uncorrelated and the variance of each variable is 1. Because this transform transforms the input noise into a white noise vector, it is called "whitening". For example, assume that X is a random noise vector with a non-singular covariance matrix Σ and a mean of 0. Then, use the condition W H W=Σ -1The transformation Y=WX of the whitening matrix W produces a whitened random vector Y with unit diagonal covariance. The output of the first noise whitening unit 404a can be a first complex channel matrix H1 associated with the first ordering of the transmission layer and a received signal Y. The first complex channel matrix H1 and Y can be input to the first QR decomposition unit 406a. The first QR decomposition unit 406a can perform QR decomposition on the first complex channel matrix H1 to generate a first estimated channel matrix and This may be input to the first tree search unit 408a.

[0130] At 608, the baseband chip may perform a first iteration of a recursive tree search operation (e.g., QR, tree search, path metric selection, etc.) on each of the plurality of first data REs for the first RB to determine a first symbol estimate associated with the first transmission layer. Figure 4 , the first complex channel matrix H1 can be input into the first QR decomposition unit 406a. The first QR decomposition unit 406a can perform QR decomposition on the first complex channel matrix H1 to generate a first estimated channel matrix and receive signals It can be input to the first tree search unit 408a. The first tree search unit 408a can be based on the first estimated channel matrix Generate a first search tree. The first tree search unit 408a can use the first search tree to perform a tree search to determine various paths and associated path metrics that can be input into the first path metric unit 410a. The technique described above in conjunction with the first iteration can be performed on each data RE in the data block received in layer 3. The first path metric unit 410a can then input information associated with each estimated data RE received across all layers into the second noise estimation unit 402b. The first path metric unit 410a can select a path with the best path metric. The path selected by the first path metric unit 410a can be associated with the estimated data RE in the data block. The first path metric unit 410a can use any hard decision technique known in the art to estimate (multiple) data REs.

[0131] At 610, the baseband chip may perform subsequent noise estimation of the channel based at least in part on the reference REs and the plurality of data REs estimated in the first iteration. Figure 4The second noise estimation unit 402b can perform channel noise estimation based at least in part on the reference REs and the estimated data REs input by the first path metric unit 410a. By using the estimated data REs and the sparse reference REs, the noise estimated by the second noise estimation unit 402b can have improved accuracy compared to conventional systems that rely solely on reference REs for noise estimation. Information associated with the second noise estimate can be input into the second noise whitening unit 404b.

[0132] At 612, the baseband chip may perform a second noise whitening operation on the channel based at least in part on the subsequent noise estimate. Figure 4 , information associated with the second noise estimate can be input to a second noise whitening unit 404b. The second noise whitening unit 404b can perform noise whitening on the second-ordered channel matrix and the received signal based at least in part on the second noise estimate. Additional details of noise whitening are provided above in conjunction with the first noise whitening unit 404a. The output of the second noise whitening unit 404b can be a second complex channel matrix H2 associated with the second ordering of the transmission layer. The second complex channel matrix H2 can be input to a second QR decomposition unit 406b.

[0133] At 614, the baseband chip may perform a second iteration of a recursive tree search operation (e.g., QR, tree search, path metric selection, etc.) on each of the plurality of data REs for the first RB to determine a second symbol estimate. Figure 4 , the second QR decomposition unit 406b may perform QR decomposition of the second complex channel matrix H2 to generate a second estimated channel matrix as well as It can be input to the second tree search unit 408b. The second tree search unit 408b can be based on the second estimated channel matrix A second search tree is generated. The second tree search unit 408b can perform a tree search using the second search tree to determine various paths and associated path metrics that can be input into the second path metric unit 410b. The second path metric unit 410b can select a path with the best path metric. The path selected by the second path metric unit 410b can be associated with the estimated data RE in the data block.

[0134] At 616, the baseband chip may generate LLRs associated with the data stream based at least in part on the first symbol estimate associated with the first transmission layer and the second symbol estimate associated with the second transmission layer. Figure 4 After performing each of the four iterations, LLR unit 410d may generate LLRs for one or more data REs. Figure 4In the described successive iterations, the baseband chip 400 can improve the noise / interference estimation by utilizing the estimated data REs in the previous iterations, compared to conventional techniques that perform noise estimation using only sparse reference REs.

[0135] In various aspects of the present disclosure, the functions described in the present disclosure can be implemented in hardware, software, firmware, or any combination thereof. If implemented in software, these functions can be stored or encoded as instructions or code on a non-transitory computer-readable medium. Computer-readable media include computer storage media. Storage media can be any available media that can be accessed by a computing device, such as Figure 7 Node 700 in. By way of example and not limitation, such computer-readable media may include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, HDD, such as magnetic disk storage or other magnetic storage device, flash drive, SSD, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and accessible by a processing system (such as a mobile device or computer). Disks and optical discs used in this disclosure include CDs, laser discs, optical discs, digital versatile discs (DVDs), and floppy disks, where disks typically reproduce data magnetically, while optical discs use lasers to reproduce data optically. Combinations of the above should also be included within the scope of computer-readable media.

[0136] According to another aspect of the present disclosure, a baseband chip including a memory and at least one processor coupled to the memory is configured to perform various operations. In some embodiments, the at least one processor is configured to receive a data stream associated with a channel. In certain aspects, the channel includes multiple transmission layers. In other certain aspects, the multiple transmission layers may include a first transmission layer associated with a first constellation point and a second transmission layer associated with a second constellation point. In some other embodiments, the at least one processor may be configured to select a first anchor point from the first constellation point of the first transmission layer. In some other embodiments, the at least one processor may be configured to select a first subset of constellation points from the first constellation point and the second constellation point based at least in part on the first anchor point. In some other embodiments, the at least one processor may be configured to perform a first iteration of a recursive tree search operation based at least in part on the first anchor point and the first subset of constellation points. In some other embodiments, the at least one processor may be configured to determine a first path metric based at least in part on the first iteration of the recursive tree search operation. In some other embodiments, the at least one processor may be configured to select a second anchor point from the second constellation point of the second transmission layer based at least in part on the first path metric. In particular aspects, the second anchor point may be associated with a second iteration of the recursive tree search operation.

[0137] In some embodiments, the at least one processor may be further configured to select a second subset of constellation points from the first constellation point and the second constellation point based at least in part on the second anchor point. In some other embodiments, the at least one processor may be further configured to perform a second iteration of the recursive tree search operation based at least in part on the second anchor point and the second subset of constellation points. In some other embodiments, the at least one processor may be further configured to determine a second path metric based at least in part on the second iteration of the recursive tree search operation.

[0138] In some embodiments, the at least one processor may be further configured to generate LLRs associated with the data flow based at least in part on the first path metric and the second path metric.

[0139] In some embodiments, the at least one processor may be further configured to perform a first iteration of a recursive tree search by performing a first ordering of the plurality of transmission layers, the first transmission layer being a bottom layer in the first ordering. In some embodiments, the at least one processor may be further configured to perform a first iteration of the recursive tree search by generating a first channel matrix based at least in part on the first ordering of the plurality of transmission layers. In some embodiments, the at least one processor may be further configured to perform a first iteration of the recursive tree search by performing a first QR decomposition of the first channel matrix. In some embodiments, the at least one processor may be further configured to perform a first iteration of the recursive tree search by performing a first tree search based at least in part on the first QR decomposition, the first anchor point, and a first subset of the constellation points.

[0140] In certain aspects, a first path metric may be determined as a result of the first tree search.

[0141] In some embodiments, the at least one processor may be further configured to perform a second iteration of the recursive tree search by performing a second ordering of the plurality of transmission layers. In certain aspects, the second transmission layer is a bottom layer in the second ordering. In some embodiments, the at least one processor may be configured to perform a second iteration of the recursive tree search by generating a second channel matrix based at least in part on the second ordering of the plurality of transmission layers. In some embodiments, the at least one processor may be further configured to perform a second iteration of the recursive tree search by performing a second QR decomposition of the second channel matrix. In some embodiments, the at least one processor may be further configured to perform a second iteration of the recursive tree search by performing a second tree search based at least in part on the second QR decomposition, the second anchor point, and a second subset of the constellation points.

[0142] In certain aspects, the first transmission layer has a higher power than the second transmission layer.

[0143] According to one aspect of the present disclosure, a baseband chip including a memory and at least one processor coupled to the memory is configured to perform various operations. In some embodiments, the at least one processor is configured to receive a data stream associated with a channel. In certain aspects, the channel may include multiple transmission layers. In other certain aspects, the multiple transmission layers may include a first transmission layer and a second transmission layer associated with a resource block (RB). In certain aspects, the RB may include at least one reference resource element (RE) and a plurality of data REs. In some other embodiments, the at least one processor is configured to perform an initial noise estimate of the channel based at least in part on the reference REs. In some other embodiments, the at least one processor is configured to perform a noise whitening operation on the channel based at least in part on the initial noise estimate. In some other embodiments, the at least one processor is configured to perform a first iteration of a recursive tree search operation on each of the plurality of data REs for the RB to determine a first symbol estimate associated with the first and second transmission layers. In some other embodiments, the at least one processor is configured to perform a subsequent noise estimate of the channel based at least in part on the reference REs, the plurality of data REs, and the first symbol estimate associated with the first iteration.

[0144] In some other embodiments, the at least one processor is configured to perform a second noise whitening operation on the channel based at least in part on the subsequent noise estimate. In some other embodiments, the at least one processor is configured to perform a second iteration of a recursive tree search operation on each of the plurality of data REs for the RB to determine a second symbol estimate associated with the second transmission layer.

[0145] In some other embodiments, the at least one processor is configured to generate LLRs associated with the data stream based at least in part on the first symbol estimate associated with the first iteration and the second symbol estimate associated with the second iteration.

[0146] In certain aspects, the first iteration of the recursive tree search operation includes a first QR decomposition and a first tree search associated with the first transmission layer. In other certain aspects, the second iteration of the recursive tree search operation includes a second QR decomposition and a second tree search associated with the second transmission layer.

[0147] According to yet another aspect of the present disclosure, a method is disclosed that may include receiving a data stream associated with a channel. In certain aspects, the channel may include multiple transmission layers. In other certain aspects, the multiple transmission layers may include a first transmission layer associated with a first constellation point and a second transmission layer associated with a second constellation point. In some other embodiments, the method may include selecting a first anchor point from the first constellation point of the first transmission layer. In some other embodiments, the method includes selecting a first subset of constellation points from the first constellation point and the second constellation point based at least in part on the first anchor point. In some other embodiments, the method may include performing a first iteration of a recursive tree search operation based at least in part on the first anchor point and the first subset of constellation points. In some other embodiments, the method may include determining a first path metric based at least in part on the first iteration of the recursive tree search operation. In some other embodiments, the method may include selecting a second anchor point from the second constellation point of the second transmission layer based at least in part on the first path metric. The second anchor point is associated with a second iteration of the recursive tree search operation.

[0148] In some other embodiments, the method may include selecting a second subset of constellation points from the first constellation point and the second constellation point based at least in part on the second anchor point. In some other embodiments, the method may include performing a second iteration of the recursive tree search operation based at least in part on the second anchor point and the second subset of constellation points. In some other embodiments, the method may include determining a second path metric based at least in part on the second iteration of the recursive tree search operation.

[0149] In some other embodiments, the method may include generating LLRs associated with the data flow based at least in part on the first path metric and the second path metric.

[0150] In some other embodiments, performing the first iteration of the recursive tree search may include performing a first ordering of the plurality of transmission layers, the first transmission layer being a bottom layer in the first ordering. In some other embodiments, the method may include generating a first channel matrix based at least in part on the first ordering of the plurality of transmission layers. In some other embodiments, performing the first iteration of the recursive tree search may include performing a first QR decomposition of the first channel matrix. In some other embodiments, performing the first iteration of the recursive tree search may include performing a first tree search based at least in part on the first QR decomposition, the first anchor point, and a first subset of the constellation points.

[0151] In some other embodiments, the method may include determining the first path metric as a result of the first tree search.

[0152] In some other embodiments, performing the second iteration of the recursive tree search may include performing a second ordering of the plurality of transmission layers, the second transmission layer being a bottom layer in the second ordering. In some other embodiments, performing the second iteration of the recursive tree search may include generating a second channel matrix based at least in part on the second ordering of the plurality of transmission layers. In some other embodiments, performing the second iteration of the recursive tree search may include performing a second QR decomposition of the second channel matrix. In some other embodiments, performing the second iteration of the recursive tree search may include performing a second tree search based at least in part on the second QR decomposition, the second anchor point, and a second subset of the constellation points.

[0153] According to another aspect of the present disclosure, a method is disclosed that may include receiving a data stream associated with a channel. In a specific aspect, the channel may include multiple transmission layers. In other specific aspects, the multiple transmission layers include a first transmission layer and a second transmission layer associated with an RB. In other specific aspects, the RB includes at least one reference RE and a plurality of data REs. In some embodiments, the method may also include performing an initial noise estimate of the channel based at least in part on the reference REs. In some embodiments, the method may also include performing a first noise whitening operation on the channel based at least in part on the initial noise estimate. In some embodiments, the method may also include performing a first iteration of a recursive tree search operation on each of the plurality of data REs for the RB to determine a first symbol estimate for the first and second transmission layers. In some embodiments, the method may also include performing a subsequent noise estimate of the channel based at least in part on the reference REs and the plurality of estimated data REs associated with the first iteration.

[0154] In some embodiments, the method may further include performing a second noise whitening operation on the channel based at least in part on the subsequent noise estimate. In some embodiments, the method may further include performing a second iteration of a recursive tree search operation on each of the plurality of data REs for the RB to determine a second symbol estimate associated with the second transmission layer.

[0155] In some embodiments, the method may further include generating LLRs associated with the data stream based at least in part on the first symbol estimate associated with the first transmission layer and the second symbol estimate associated with the second transmission layer.

[0156] The above description of specific embodiments will reveal the general nature of the present disclosure, and others can easily modify and / or adjust these specific embodiments for various applications by applying knowledge within the scope of the art without the need for excessive experimentation or departure from the general concept of the present disclosure. Therefore, based on the teachings and guidance presented herein, such modifications and adjustments are intended to be within the meaning and scope of the equivalents of the disclosed embodiments. It should be understood that the wording or terminology herein is for descriptive and not limiting purposes, so that the terms or wording of this specification will be interpreted by those skilled in the art based on the teachings and guidance.

[0157] The embodiments of the present disclosure have been described above with the aid of functional blocks that illustrate the implementation of specific functions and their relationships. For ease of description, the boundaries of these functional blocks are arbitrarily defined herein. Alternative boundaries may be defined as long as the specified functions and relationships are properly performed.

[0158] The Summary and Abstract sections may set forth one or more but not all exemplary embodiments of the present disclosure as contemplated by the inventor(s), and thus, are not intended to limit the present disclosure and the appended claims in any way.

[0159] Various functional blocks, modules, and steps are disclosed above. The specific arrangements provided are illustrative and not limiting. Therefore, the functional blocks, modules, and steps may be reordered or combined in a manner different from the examples provided above. Similarly, some embodiments include only a subset of the functional blocks, modules, and steps, and any such subset is permitted.

[0160] The breadth and scope of the present disclosure should not be limited by any of the above-described exemplary embodiments, but should be defined only in accordance with the following claims and their equivalents.

Claims

1. A baseband chip for wireless communication, comprising: Memory; as well as at least one processor coupled to the memory and configured to: receiving a data stream associated with a channel, the channel comprising a plurality of transmission layers, the plurality of transmission layers comprising a first transmission layer associated with a first constellation point and a second transmission layer associated with a second constellation point; selecting a first anchor point from the first constellation points of the first transmission layer; selecting a first subset of constellation points from the first constellation points and the second constellation points based at least in part on the first anchor point; performing a first iteration of a recursive tree search operation based at least in part on the first anchor point and the first subset of constellation points; determining a first path metric based at least in part on the first iteration of the recursive tree search operation; as well as A second anchor point is selected from the second constellation points of the second transmission layer based at least in part on the first path metric, the second anchor point being associated with a second iteration of the recursive tree search operation.

2. The baseband chip according to claim 1, wherein: The at least one processor is further configured to: selecting a second subset of constellation points from the first constellation point and the second constellation point based at least in part on the second anchor point; performing a second iteration of the recursive tree search operation based at least in part on the second anchor point and the second subset of constellation points; as well as A second path metric is determined based at least in part on the second iteration of the recursive tree search operation.

3. The baseband chip according to claim 2, wherein: The at least one processor is further configured to: A log-likelihood ratio (LLR) associated with the data stream is generated based at least in part on the first path metric and the second path metric.

4. The baseband chip according to claim 1 , wherein the at least one processor is configured to perform the first iteration of the recursive tree search operation by: performing a first ordering of the plurality of transport layers, the first transport layer being a bottom layer in the first ordering; generating a first channel matrix based at least in part on the first ordering of the plurality of transmission layers; performing a first QR decomposition of the first channel matrix; as well as A first tree search is performed based at least in part on the first QR decomposition, the first anchor point, and the first subset of constellation points. The baseband chip according to claim 4 , wherein the first path metric is determined as a result of the first tree search.

6. The baseband chip according to claim 2, wherein the at least one processor is configured to perform the second iteration of the recursive tree search operation by: performing a second ordering of the plurality of transport layers, the second transport layer being a bottom layer in the second ordering; generating a second channel matrix based at least in part on the second ordering of the plurality of transmission layers; performing a second QR decomposition of the second channel matrix; as well as A second tree search is performed based at least in part on the second QR decomposition, the second anchor point, and the second subset of constellation points.

7. The baseband chip according to claim 1, wherein: The power of the first transmission layer is higher than that of the second transmission layer.

8. A baseband chip for wireless communication, comprising: Memory; as well as at least one processor coupled to the memory and configured to: receiving a data stream associated with a channel, the channel comprising a plurality of transmission layers, the plurality of transmission layers comprising a first transmission layer and a second transmission layer associated with a resource block (RB), the RB comprising at least one reference resource element (RE) and a plurality of data REs; performing an initial noise estimation of the channel based at least in part on the at least one reference RE; performing a first noise whitening operation on the channel based at least in part on the initial noise estimate; performing a first iteration of a recursive tree search operation on each of the plurality of data REs for the RB to determine a first symbol estimate associated with the first transmission layer; and A subsequent noise estimation of the channel is performed based at least in part on the at least one reference RE, the plurality of data REs, and the first symbol estimate associated with a first iteration of the recursive tree search.

9. The baseband chip according to claim 8, wherein: The at least one processor is further configured to: performing a second noise whitening operation on the channel based at least in part on the subsequent noise estimate; as well as A second iteration of a recursive tree search operation is performed on each of the plurality of data REs for the RB to determine a second symbol estimate associated with the second transmission layer.

10. The baseband chip according to claim 9, wherein: The at least one processor is further configured to: Log-likelihood ratios (LLRs) associated with the data stream are generated based at least in part on the first symbol estimate associated with the first iteration and the second symbol estimate associated with the second iteration.

11. The baseband chip according to claim 10, wherein: The first iteration of the recursive tree search operation includes a first QR decomposition and a first tree search associated with the first transport layer, and The second iteration of the recursive tree search operation includes a second QR decomposition and a second tree search associated with the second transmission layer.

12. A wireless communication method, comprising: receiving a data stream associated with a channel, the channel comprising a plurality of transmission layers, the plurality of transmission layers comprising a first transmission layer associated with a first constellation point and a second transmission layer associated with a second constellation point; selecting a first anchor point from the first constellation points of the first transmission layer; selecting a first subset of constellation points from the first constellation points and the second constellation points based at least in part on the first anchor point; performing a first iteration of a recursive tree search operation based at least in part on the first anchor point and the first subset of constellation points; determining a first path metric based at least in part on the first iteration of the recursive tree search operation; as well as A second anchor point is selected from the second constellation points of the second transmission layer based at least in part on the first path metric, the second anchor point being associated with a second iteration of the recursive tree search operation.

13. The method according to claim 12, further comprising: selecting a second subset of constellation points from the first constellation point and the second constellation point based at least in part on the second anchor point; performing a second iteration of the recursive tree search operation based at least in part on the second anchor point and the second subset of constellation points; as well as A second path metric is determined based at least in part on the second iteration of the recursive tree search operation.

14. The method according to claim 13, further comprising: A log-likelihood ratio (LLR) associated with the data stream is generated based at least in part on the first path metric and the second path metric.

15. The method of claim 12, wherein said performing said first iteration of a recursive tree search operation comprises: performing a first ordering of the plurality of transport layers, the first transport layer being a bottom layer in the first ordering; generating a first channel matrix based at least in part on the first ordering of the plurality of transmission layers; performing a first QR decomposition of the first channel matrix; as well as A first tree search is performed based at least in part on the first QR decomposition, the first anchor point, and the first subset of constellation points. The method of claim 15 , wherein the first path metric is determined as a result of the first tree search.

17. The method of claim 13, wherein said performing said second iteration of a recursive tree search operation comprises: performing a second ordering of the plurality of transport layers, the second transport layer being a bottom layer in the second ordering; generating a second channel matrix based at least in part on the second ordering of the plurality of transmission layers; performing a second QR decomposition of the second channel matrix; as well as A second tree search is performed based at least in part on the second QR decomposition, the second anchor point, and the second subset of constellation points.

18. A method of wireless communication, comprising: receiving a data stream associated with a channel, the channel comprising a plurality of transmission layers, the plurality of transmission layers comprising a first transmission layer and a second transmission layer associated with a resource block (RB), the RB comprising at least one reference resource element (RE) and a plurality of data REs; performing an initial noise estimation of the channel based at least in part on the at least one reference RE; performing a first noise whitening operation on the channel based at least in part on the initial noise estimate; performing a first iteration of a recursive tree search operation on each of the plurality of data REs for the RB to determine a first symbol estimate associated with the first transmission layer; and A subsequent noise estimation of the channel is performed based at least in part on the at least one reference RE, the plurality of data REs, and the first symbol estimate associated with a first iteration of the recursive tree search.

19. The method according to claim 18, further comprising: performing a second noise whitening operation on the channel based at least in part on the subsequent noise estimate; as well as A second iteration of a recursive tree search operation is performed on each of the plurality of data REs for the RB to determine a second symbol estimate associated with the second transmission layer.

20. The method according to claim 19, further comprising: Log-likelihood ratios (LLRs) associated with the data stream are generated based at least in part on the first symbol estimate associated with the first iteration and the second symbol estimate associated with the second iteration.

Images