A low-latency adaptive configuration method for multi-hop RIS-assisted communication

Abstract

The application discloses a low-latency adaptive configuration method for multi-hop RIS-assisted communication, which is based on a network graph model. Under the condition of meeting the constraints of receiving power threshold and maximum hop number, multiple feasible multi-hop candidate paths are generated for each user. Then, a candidate path conflict graph is constructed, and paths that share RIS nodes or exist resource contention are marked as conflict edges. The maximum independent set is solved to form a user group that can be concurrently activated. At the RIS hardware configuration layer, the proportion and number of active units of each RIS are adaptively determined by using a smooth interpolation method. Furthermore, the RIS units are greedily selected by Top-K selection according to the end-to-end gain contribution, so as to realize efficient deployment of hybrid active / passive RIS. At the physical layer, an end-to-end equivalent channel model is constructed based on the selected path and RIS configuration, and low-complexity beamforming within the group and time slot proportion allocation between groups are jointly performed to improve the minimum user average reachable rate of the system.

Description

Technical Field

[0001] This invention belongs to the field of communication technology, specifically relating to a low-latency adaptive configuration method for multi-hop RIS-assisted communication, which can be applied to 6G mobile communication networks that are oriented towards ultra-dense access, enhanced edge coverage, and high-dynamic services. Background Technology

[0002] As the vision for sixth-generation (6G) mobile communication systems becomes clearer, network objectives are shifting from "improving peak rates on single links" to "ensuring deterministic user experience in multi-user scenarios," with particular emphasis on maintaining stable throughput and low-latency interaction under conditions of complex urban obstruction, sudden service disruptions, and user mobility. To meet these requirements, massive MIMO and beamforming technologies are widely used to improve link gain, but their performance is highly dependent on the network's real-time understanding of topology and channel conditions. In ultra-dense access and dynamic service scenarios, a single base station or a single RIS (Real-Risk Array) can hardly simultaneously ensure coverage continuity, anti-obstruction capability, and multi-user fairness. Therefore, multi-RIS collaboration, or even multi-hop relay-style RIS link construction, has become a potential means to improve edge coverage and link reliability.

[0003] However, the introduction of multiple RIS and multi-hop links also brings significant systemic challenges. First, the number of candidate paths increases exponentially with the number of RIS nodes and hops, and multi-user parallel optimization can lead to an explosion in the search space, making it difficult to meet online scheduling latency requirements. Second, multi-user paths often generate competition and interference on RIS nodes or spatial neighborhoods. Without systematic conflict modeling and group scheduling mechanisms, concurrent transmission can significantly degrade link quality and even cause "local optima but global instability." Third, although hybrid active / passive RIS can compensate for the insufficient gain of pure passive reflection to some extent, active units introduce power consumption and noise amplification. If a fixed ratio or random selection of active units is used, a dilemma of "either excessive power consumption or insufficient gain" often occurs under different user distances and topology loads. Finally, even if precoding and time slot allocation optimization are performed under given path and RIS configuration conditions, global re-optimization at every moment will bring unacceptable computational overhead, while long-term reuse of old solutions will lead to rapid performance degradation due to channel evolution.

[0004] Therefore, existing technologies urgently need a method for multi-RIS multi-hop mobile communication that can achieve concurrent group scheduling, efficient configuration of RIS active units under power constraints, and low-latency dynamic updates while ensuring fairness and throughput for multiple users. This would enable multi-RIS multi-hop networks to have real-time optimization capabilities that can be deployed in engineering. Summary of the Invention

[0005] The purpose of this invention is to provide a low-latency adaptive configuration method for multi-hop RIS-assisted communication, which aims to solve problems such as huge candidate path search space, strong mutual interference of concurrent transmission, limited energy consumption and difficult configuration of hybrid active RIS in multi-RIS multi-hop scenarios, and excessively high latency of online optimization under dynamic channels, thereby ensuring the continuous availability of links, system fairness and spectrum efficiency in mobile scenarios.

[0006] To achieve the above objectives, the present invention employs the following technical solution: A low-latency adaptive configuration method for multi-hop RIS-assisted communication includes: In a network consisting of a base station, multiple active / passive RIS nodes, and multiple users, a candidate path set containing at least one RIS node is generated for each user. During the generation process, maximum hop count and received power threshold constraints are introduced to control complexity. A path conflict graph is constructed, and the shared RIS nodes or path pairs with resource contention between multiple user paths are explicitly modeled as conflict edges in a graph structure. By solving for the maximum independent set or the equivalent set of conflict-free path groups, concurrently activated path groups are obtained. For each hybrid active / passive RIS node, based on the user distance information corresponding to the path served by the RIS node, a smooth interpolation mechanism is used between the near distance threshold and the far distance threshold to adaptively determine the proportion of active units in each RIS node, so that the proportion of active units changes continuously with the network structure and service load. After determining the proportion of active units, a contribution index is constructed for all units within each hybrid active / passive RIS node. This contribution index reflects the joint gain of the unit in the two adjacent links in the selected path, and is weighted and accumulated by combining the large-scale fading coefficients of the two links. Then, units are sorted according to the contribution index, and a preset number of units are selected as the active unit set. The group-activated multi-hop links are mapped to a group-level equivalent channel model, thereby converting the combined effect of multi-hop links on users into an equivalent expression that can be used for joint optimization. On this group-level equivalent channel model, the allocation optimization of joint precoding vectors and time slot ratios is performed to maximize the minimum average reachability of the system.

[0007] Furthermore, the link is evolved using a Jakes or Gauss-Markov time correlation model, and joint optimization is performed according to a preset update cycle. During non-update times, the allocation results of concurrently activated path groups, precoding vectors, and time slot ratios obtained in the previous cycle are reused.

[0008] Furthermore, a candidate path set containing at least one RIS node is generated for each user, and maximum hop count and received power threshold constraints are introduced during the generation process to control complexity, including: The path generation problem for each user is transformed into a constrained shortest path search problem: for each user... k On the node graph, from the base station node To user node Search candidate path set And apply maximum hop count and receive power threshold constraints: ; ; in, Indicates selection The smallest front K Path; Representing paths The starting node and any node; Indicates the path direction; For path The number of jumps; The maximum number of hops allowed; This is the reachability threshold for the received power; For path The cumulative cost of losses; To estimate the received power: ; in, This refers to the base station's transmission power. For path Path loss, This represents the power gain of a single GIS node.

[0009] Furthermore, the process of constructing the path conflict graph is as follows: Let the set of RIS nodes used by any path be denoted as . Two paths are considered to be in conflict if either of them satisfies one of the following conditions: Shared RIS nodes; RIS node clusters are spatially adjacent; Therefore, a conflict determination function is constructed. Conflict determination function A value of 1 indicates a path Conflict, 0 indicates no conflict; Construct a path conflict graph ;in Let the set of edges be defined as follows: if the paths between two vertices are... Conflict conditions are met If two paths cannot be activated simultaneously, then an edge is drawn between them; vertex set It consists of "user-path" pairs, where each vertex represents a candidate service option. It can be written as: ; in, For user collection; For users k The set of candidate paths; if the two paths corresponding to two vertices conflict, then connect them with an edge.

[0010] Furthermore, by solving for the maximum independent set or the equivalent set of conflict-free path groups, concurrently activated path groups are obtained, including: In a path conflict graph, a concurrently activated path group must satisfy the following condition: any two vertices within the group... There are no conflicting edges between them; let the first... g The concurrent activation path groups are denoted as follows: Then the independence constraint is: ; in, Represents two vertices There is no edge between them; Concurrent activation path grouping is constructed using the maximum independent set criterion: ; in, Represents the vertex set A subset of; This represents a function that finds the index of the maximum value.

[0011] Furthermore, a smooth interpolation mechanism is employed between the near-distance threshold and the far-distance threshold to adaptively determine the proportion of active units in each RIS node, including: Let the set of service users of the r-th RIS node be . The service distance metric is defined as the average distance: ; in, Let be the distance from the r-th RIS node to user k; A smooth interpolation function is used to map the distance to the proportion of active cells in the r-th RIS node. : ; in, , These are the pre-set near-distance threshold and far-distance threshold, respectively; , The pre-set ratio reference values ​​for near-end active nodes and far-end active nodes; Therefore, the number of active units that need to be activated in the r-th RIS node is: ; in This indicates rounding down to the nearest integer.

[0012] Furthermore, a contribution index is constructed for all units within each hybrid active / passive RIS node; a preset number of units are selected as the active unit set, including: For cell n of the r-th RIS node, let the preceding and following nodes adjacent to the r-th RIS node in the path be respectively... and In the currently selected service path set Above, define the contribution index: ; in , For the node Large-scale fading coefficient and incident channel component up to the r-th RIS node; , For the r-th RIS node to node Large-scale fading coefficient and outgoing channel components; The set of active cells is determined using a greedy Top-K strategy: ; in, This indicates the selection of the indices of the K active units with the largest contribution indicators; and the amplitude coefficients are... The value assigned is: ; in, This represents the amplitude gain of the active unit.

[0013] Furthermore, for a certain RIS node on path p The local cascade complex coefficients of its nth unit are as follows: ; in and These represent the previous and next hop nodes of the RIS in the path, respectively; They represent To RIS node RIS node arrive The incident channel component and the outgoing channel component; The phase control of this unit adopts the conjugate alignment principle: ; in, Represents RIS node The phase value of the nth unit on path p; This indicates taking the phase angle; Therefore, the configuration matrix of RIS node r The nth diagonal element It can be written as: ; in This is the final configuration of the nth unit of RIS node r; j It is the imaginary unit.

[0014] Furthermore, the group-level equivalent channel model is constructed as follows: ; Joint optimization of the group-level equivalent channel model yields the precoding vector. and time slot ratio ; This represents the maximum total transmit power of the base station. express l 2-norm; Let the time slot ratio occupied by the g-th concurrent activation path group be... Then the following conditions are met: ; Where G is the total number of concurrently activated path groups; The average reachability of user k is: ; Let g be the set of service users within the g-th concurrent activation path group; Intra-group instantaneous spectral efficiency: ; in Let S be the signal-to-interference-plus-noise ratio (SIR) for user k within the concurrent active path group.

[0015] A terminal device includes a processor, a memory, and a computer program stored in the memory; when the processor executes the computer program, it implements the low-latency adaptive configuration method for multi-hop RIS-assisted communication.

[0016] A computer-readable storage medium storing a computer program; when executed by a processor, the computer program implements the low-latency adaptive configuration method for multi-hop RIS-assisted communication.

[0017] Compared with the prior art, the present invention has the following technical features: 1. Fundamentally reduce online optimization overhead and latency: Compared with traditional schemes that perform a one-time global exhaustive search or large-scale joint iteration of "path selection - RIS configuration - beamforming - resource allocation" across the entire network, this invention adopts a hierarchical structure of "candidate path generation, conflict graph grouping, and intra-group joint optimization" to decompose the originally exponentially expanding search space into multiple low-dimensional subproblems that can be processed in parallel. At the same time, by introducing a mechanism of periodic re-optimization and reuse of old solutions at intermediate moments, rate evaluation and link tracking can be completed without performing complete joint optimization most of the time, thereby significantly reducing system-level online computation latency and making it feasible for real-time implementation in dynamic mobile scenarios.

[0018] 2. Achieving stable concurrency and fair rate guarantee in multi-user, multi-RIS, and multi-hop scenarios: This invention obtains a set of concurrently active groups by constructing a path conflict graph and solving the maximum independent set (MIS), thus structurally avoiding performance collapse caused by shared RIS / strong mutual interference in multi-user concurrent transmission; furthermore, joint precoding and time slot allocation are performed within each concurrent group, directly optimizing based on the minimum user rate or fairness index, thereby avoiding the phenomenon of "strong users monopolizing resources and weak users experiencing rate collapse", significantly improving multi-user fairness and system stability.

[0019] 3. High-efficiency gain release of hybrid active RIS under power constraints: To address the configuration challenges of hybrid active / passive RIS, this invention introduces an active ratio decision mechanism that adapts to link geometry / service distance, allowing the number of active units to dynamically change with scenario requirements, avoiding energy waste or insufficient gain caused by a fixed ratio; and proposes a RIS unit selection criterion based on "joint contribution of the two links before and after the path", prioritizing the activation of units that contribute the most to the end-to-end equivalent link, achieving higher equivalent channel gain and better rate improvement under the same active budget, thereby achieving a balance between energy efficiency and performance.

[0020] 4. Forming a closed-loop system of "network structure decision-making - physical layer joint optimization - performance evaluation": This invention tightly couples multi-hop path and packet scheduling (network layer structure) with precoding / time slot allocation under equivalent channels (physical layer optimization), and uses system-level indicators (minimum rate, average rate, total throughput, etc.) as feedback to construct a complete structured closed-loop optimization framework. This closed loop not only quantifies the impact of different packets and different RIS active configurations on communication performance, but also provides a basis for dynamic update triggering, thereby achieving interpretable and controllable performance evolution under complex topologies and dynamic channels.

[0021] 5. Improve robustness in dynamic scenarios and reduce performance fluctuations caused by channel evolution: Under mobile / time-varying channel conditions, this invention adopts a time-related channel evolution model to characterize link changes, and through the strategy of "intermittent re-optimization + intermediate step fast tracking", the system can update path grouping and RIS configuration at critical moments to adapt to environmental changes, and maintain structural stability and quickly refresh performance evaluation results most of the time, thereby reducing instability and overhead caused by frequent switching and improving link robustness and throughput performance in dynamic scenarios. Attached Figure Description

[0022] Figure 1 This is a system block diagram of the method of the present invention; Figure 2 This is a schematic flowchart of the method of the present invention; Figure 3 The communication rate of the method of the present invention under different active RIS ratios; Figure 4 Comparison of the speed of the method of the present invention and the MRT algorithm under different active RIS ratios; Figure 5 The minimum rate, average rate, and fairness gain of the method of the present invention relative to the MRT are given under different modes; Figure 6 The curves show the increase in minimum and average rates for mobile users using the method of the present invention.

[0023] The method of the present invention is referred to as Approx-SDP in the accompanying drawings. Detailed Implementation

[0024] As immersive extended reality (XR), holographic communication, vehicle-to-everything (V2X) and low-altitude mobile access services place increasingly stringent demands on "high throughput, low latency, and high reliability," the network side not only needs to provide high-gain beam coverage but also needs to achieve stable fair rate guarantees and low-complexity online optimization capabilities under multi-user, multi-node collaborative conditions. This invention focuses on a mobile communication system assisted by a reconfigurable intelligent surface (RIS). By introducing multiple RIS, multi-hop links, and concurrent packet scheduling mechanisms, it solves the problems of path selection, resource conflicts, and available gain acquisition under energy-constrained conditions in complex topologies, thereby improving end-to-end connection robustness and spectral efficiency in dynamic scenarios.

[0025] This invention provides a low-latency adaptive configuration method for multi-hop RIS-assisted communication, constructing an integrated structure that spans "path-grouping-RIS configuration-resource allocation-dynamic update," enabling the system to achieve stable operation through hierarchical decision-making even in complex topologies. The method includes the following steps: Step 1: In a network consisting of a base station, multiple active / passive RIS nodes, and multiple users, a candidate path set containing at least one RIS node is generated for each user. During the generation process, maximum hop count and received power threshold constraints are introduced to control complexity. Based on this, a path conflict graph is constructed, and the shared RIS nodes or path pairs with resource contention between multiple user paths are explicitly modeled as conflict edges in a graph structure. By solving the maximum independent set (MIS) or the equivalent set of conflict-free path groups, concurrent active path groups are obtained, ensuring the feasibility and stability of concurrent transmission from the structural layer.

[0026] The path grouping mechanism described above decomposes the originally heavily coupled multi-user parallel optimization problem into several low-conflict activation groups, enabling the system to suppress the spread of mutual interference while ensuring concurrency efficiency, and providing stable structural constraints for subsequent RIS configuration and resource allocation.

[0027] Step 2: On the RIS node side, for each hybrid active / passive RIS node, instead of using a fixed ratio or random activation method, the proportion of active units in each RIS node is adaptively determined based on the user distance statistics or link scale information corresponding to the path served by the RIS node. A smooth interpolation mechanism is used between the near-distance threshold and the far-distance threshold to make the proportion of active units continuously change with network geometry and service load. This allows for the adaptive allocation of the number of active units in different locations, enabling a controllable trade-off between "gain-power consumption-noise amplification" in different scenarios.

[0028] Step 3: After determining the proportion of active units, a contribution index is constructed for all units (including active and passive units) within each hybrid active / passive RIS node. This contribution index reflects the joint gain of the unit in the two adjacent links (from the previous hop to the RIS node and from the RIS node to the next hop) in the selected path, and is weighted and accumulated by combining the large-scale fading coefficients of the two links. Then, the units are sorted according to the contribution index, and a preset number of units are selected as the active unit set in a greedy manner.

[0029] By using this "front-end and back-end link coupling" scoring and greedy selection, the present invention can prioritize the activation of the unit that contributes the most to the end-to-end gain under a given active number constraint, which significantly improves the energy efficiency and link gain release effect of the hybrid active / passive RIS node.

[0030] Step 4: In terms of physical layer resource allocation, the group-activated multi-hop links are mapped to a group-level equivalent channel model, thereby converting the comprehensive effect of multi-hop links on users into an equivalent expression that can be used for joint optimization; on this group-level equivalent channel model, the allocation optimization of joint precoding vectors and time slot ratios is performed to maximize the minimum average reachability rate of the system.

[0031] This joint optimization is coupled with the aforementioned grouping structure and RIS greedy configuration: grouping reduces mutual interference, making the equivalent model more stable, and the RIS greedy configuration reduces the optimization dimension and improves the quality of the equivalent channel, so that the system can still obtain considerable performance gains under complex topologies.

[0032] Based on the above scheme, in order to meet the real-time requirements of mobile scenarios, a dynamic low-latency update mechanism is introduced: the link is evolved using a Jakes or Gauss-Markov time correlation model, and the allocation optimization in step 4 above is performed according to a preset update cycle; the allocation results of path grouping, precoding vector and time slot ratio obtained in the previous cycle are reused at non-update times; the rate index is quickly evaluated on the updated equivalent channel to realize the online operation mode of "periodic re-optimization + intermediate step reuse of old solution for fast tracking". While ensuring that the system performance does not significantly degrade, the online computing overhead is significantly reduced, enabling the multi-RIS node multi-hop network to have an engineering-deployable low-latency adaptive capability.

[0033] In summary, this invention organically integrates multi-hop concurrent group scheduling, hybrid active proportional adaptive scheduling and RIS unit adaptive selection, joint precoding and time slot allocation under equivalent channels, and a dynamic low-latency update mechanism to form a holistic structured solution for multi-RIS multi-hop mobile communication, reducing online optimization latency and computational complexity. Compared with traditional single-hop RIS or conflict-free scheduling schemes, this invention can significantly improve concurrency, user fairness, and dynamic robustness while ensuring controllable online complexity, making it suitable for latency-sensitive services such as immersive XR and vehicle-to-everything (V2X) in 6G mobile communication.

[0034] Example

[0035] Implementation Step 1: Parametric network link construction is a method that leverages the inherent physical geometry and large-scale fading patterns of communication networks to significantly reduce the complexity of online decision-making. The core idea of ​​this method differs fundamentally from the traditional approach of "joint optimization across the entire channel matrix domain": it does not attempt to recover or optimize a high-dimensional object with massive degrees of freedom at each time step (e.g., all RIS unit phases, all multi-hop combinations, and global joint variables of all users). Instead, it focuses on extracting and searching a few structural parameters that determine end-to-end link reachability and coarse gain levels, thus transforming online computation from "continuous high-dimensional optimization" to "structured search of a finite set." This approach is particularly suitable for dynamic networks with a large number of RIS units, rapid user movement, and frequent reconfiguration requirements.

[0036] The core idea of ​​the method proposed in this invention is to abstract the multi-hop communication network of "base station-multiple RIS-user" into a weighted graph structure composed of nodes and links. Link weights are constructed using physical quantities such as distance between nodes and carrier frequency, thereby reconstructing the originally complex multi-hop cascaded channel problem into a parameterized search problem dominated by path sequences in the first stage. In other words, instead of directly optimizing the phase shift matrix of all RIS units in step 1... Compared with the pre-encoding W within the group, this step only solves the problem of "which multi-hop path is more feasible and more worthy of subsequent fine optimization", extracting the dominant factors of the network structure as inputs for subsequent grouping (MIS) and RIS greedy selection of units.

[0037] To achieve the aforementioned "parameterized path search," this invention first models the large-scale loss of any two nodes i and j, and maps it to edge weights on the graph. For a distance of... The path loss (dB) of the link can be written as:

[0038] in This is the carrier frequency. The loss is then normalized to form an additive weight:

[0039] Thus, any path The cumulative cost of loss can be expressed as: ,

[0040] in The jump number is the sequence number; For nodes The weights between them; For path Path loss.

[0041] In multi-hop RIS links, RIS reflections tend to lead to a cumulative cascading gain. To enable a rapid assessment of whether multiple hops are worthwhile during the path search phase, this invention introduces a reference hybrid gain factor in the path feasibility determination to roughly characterize the "average equivalent gain per RIS node". Let the source cell amplitude gain be... The amplitude of the passive unit is The reference active ratio is Then the mixing amplitude factor of a single RIS node can be written as: ,

[0042] in Power gain for a single GIS node; in the pathp Above, if its jump count is L , which includes L With -1 RIS node participation (i.e., excluding the starting BS and ending UE), the estimated received power is expressed as:

[0043] in This refers to the base station's transmission power. For path Path loss.

[0044] The significance of this expression is that, before entering the small-scale channel and fine-tuning the RIS phase, the path loss and multi-hop gain compensation are used to quickly determine whether the long path is still reachable.

[0045] Based on the above parameterized model, this invention transforms the path generation problem for each user into a constrained shortest path search problem: for each user k From the base station node on the node graph To user node Search candidate path set And apply maximum hop count and receive power threshold constraints:

[0046]

[0047] in, Indicates selection The smallest front K Path; For path The cumulative cost of loss; Representing paths The starting node and any node; Indicates the path direction; For path The number of jumps; The maximum number of hops allowed; This is the reachability threshold for received power (corresponding to dB).

[0048] In this way, the present invention transforms the first stage of "multi-RIS multi-hop cascaded channel optimization" from a high-dimensional continuous problem into a search that is only related to a small number of discrete variables: each user only needs to determine the node sequence of a finite number of candidate paths.

[0049] Compared with traditional direct joint optimization based on the global channel matrix, parameterized path generation based on physical structure has the following significant advantages: (1) Fundamentally reduce search size: This step does not require enumerating all RIS phases and all multi-hop combinations, but instead limits the candidate set for each user to a smaller number of RIS phases using K-short circuits. The limited scale avoids combinatorial explosion, ensuring that subsequent grouping and beam optimization are only performed on "a few worthwhile paths".

[0050] (2) Strong physical interpretability: The quality of the path is determined by distance, carrier frequency, loss and reference hybrid gain, which has clear physical meaning and avoids the problem of black box decision-making being difficult to adjust parameters and difficult to reproduce stably in dynamic scenarios.

[0051] (3) Suitable for rapid updates in dynamic scenarios: When user movement causes changes in geometric distance, the weight matrix... The reachability determination can be quickly recalculated, enabling the candidate path set to track topology changes with low overhead, providing a basis for low-latency reconfiguration.

[0052] (4) Provide structural priors for subsequent selection of active units: The candidate path sequence given in this step directly determines the "previous hop / next hop" relationship of each RIS in the path, so that the unit scoring function of "joint contribution of the two segments" can be constructed in the future, and the active budget can be used efficiently.

[0053] It should be noted that although this step significantly reduces complexity through parameterized path search, in multi-user concurrent service scenarios, the paths chosen by different users may share RIS nodes or be spatially highly proximate, leading to resource conflicts and mutual interference coupling. Without structured processing, even with fine beam optimization for each path, system instability may still result from concurrent conflicts. Therefore, it is necessary to further design conflict graph modeling and a maximum independent set (MIS) grouping mechanism to address the concurrency feasibility issue, which is the core issue that subsequent implementation steps will address.

[0054] Implementation Step 2: In multi-user, multi-RIS, multi-hop systems, a core challenge lies not in whether a single path is "reachable," but in whether multiple paths can be activated concurrently. This is because candidate paths from different users often share the same RIS node or are carried by spatially adjacent RIS clusters, leading to strong mutual interference, reflection resource conflicts, and the accumulation of control overhead. Traditional approaches often address mutual interference coupling only during the physical layer joint optimization phase, resulting in extremely high dimensionality of optimization variables, unstable solutions, and difficulty in meeting the low-latency requirements of dynamic scenarios.

[0055] To address the aforementioned issues, this invention introduces a structured mechanism called "Conflict Graph Modeling - Maximum Independent Set Grouping (MIS)" after candidate path generation. This mechanism first explicitly defines which paths cannot be opened simultaneously, then transforms concurrent scheduling from a complex continuous optimization problem into a discrete graph selection problem. Essentially, this step reconstructs the feasibility of multi-user concurrency from a complex constraint "implicit in channel coupling" to a "mutually exclusive edge constraint on the graph," thereby significantly reducing the difficulty of subsequent joint precoding and RIS configuration and improving system-level stability.

[0056] (1) Formal definition of path conflict relationship.

[0057] Let the set of RIS nodes used by any path be denoted as . Two paths are considered to be in conflict if either of them satisfies one of the following conditions: (i) Shared RIS nodes (the same RIS node cannot serve simultaneously or may lead to strong coupling); (ii) Spatial proximity of RIS node clusters (concurrency of neighboring RIS can cause significant interference or control coupling).

[0058] The conflict determination function can be written as:

[0059] in Paths The set of RIS nodes used; RIS nodes r , s They are respectively Nodes in; For RIS nodes r , s Spatial distance; The nearest collision threshold; the collision determination function. A value of 1 indicates a path Conflict, 0 indicates no conflict.

[0060] (2) Path conflict graph construction.

[0061] This invention constructs a path conflict graph. ;in Let the set of edges be defined as follows: if the paths between two vertices are... Conflict conditions are met If two paths cannot be activated simultaneously, then an edge is drawn between them; vertex set It consists of "user-path" pairs, where each vertex represents a candidate service option. It can be written as:

[0062] Among them, candidate service selection Characterizing users k A candidate path p ; For user collection; For users k The set of candidate paths; if two paths corresponding to two vertices conflict, then connect them with an edge:

[0063] in, For two different users.

[0064] Through the above modeling, the system transforms the question of "whether concurrency is possible" into mutual exclusion edge constraints on the graph, providing clear structural constraint inputs for subsequent solutions to concurrent grouping.

[0065] (3) Maximum Independent Set (MIS) grouping: obtain the set of paths that can be activated concurrently.

[0066] In a path conflict graph, a concurrently activated path group must satisfy the following condition: any two vertices within the group... There are no conflicting edges between them, meaning the group is an independent set; let the first... g The concurrent activation path groups are denoted as follows: Then the independence constraint is:

[0067] in, Represents two vertices There is no edge between them.

[0068] To maximize concurrency, this invention employs the Maximum Independent Set (MIS) criterion to construct concurrent activation path groups (or uses an approximation algorithm to approximate the MIS):

[0069] in, Represents the vertex set A subset of.

[0070] For ease of engineering implementation, this problem can also be written in 0-1 integer programming form. Let each vertex... Corresponding to a binary choice variable ,but:

[0071] At the code implementation level, this invention can use a greedy / heuristic approach to find an approximate MIS (e.g., adding nodes one by one according to their degree or path "potential benefit") to obtain low-latency concurrent activation path grouping results.

[0072] (4) Grouping sequence generation.

[0073] After obtaining a concurrent activation path group, in order to cover the service needs of all users, this invention generates multiple sets of concurrent activation path groups using an "iterative stripping" method: each time an independent set is obtained. Then, the vertices contained in it are removed from the candidate set, and the solution is repeated for the remaining vertices to finally obtain the concurrently activated path group set. .

[0074] This step addresses the structural problem of "who can serve simultaneously and which path to use for concurrency," but it doesn't determine the internal configuration of the RIS, especially in a hybrid active / passive RIS setup. It addresses how to select the active ratio and place active cells in the "most valuable positions" under power constraints to maximize end-to-end equivalent gain and improve intra-group speed. Therefore, it's necessary to further design an adaptive active ratio + greedy RIS cell selection mechanism and couple it with the path's forward and backward hop structure. This is the core issue that subsequent implementation steps will address.

[0075] Implementation Step 3: In hybrid active / passive RIS node-assisted communication systems, the configurable degrees of freedom of RIS nodes derive from the amplitude and phase control of a large number of cells. If the amplitude and phase of each cell are directly optimized globally and continuously, the dimension of variables will increase linearly with the RIS scale, resulting in extremely high online solution overhead and difficulty in meeting low latency requirements in mobile scenarios. More importantly, while active cells can provide gain, they also introduce additional power consumption and noise amplification risks. If active cells are activated at fixed ratios or random locations, the phenomenon of "active budget exhaustion with limited end-to-end gain improvement" often occurs, making it difficult to balance system energy efficiency and performance.

[0076] To address the aforementioned contradictions, this invention proposes an integrated configuration method combining active ratio adaptive optimization and greedy selection of cell contribution. Its core idea is: instead of directly optimizing the continuous variables of all cells in step 2, the configuration problem of RIS nodes is first decomposed into two levels: Proportional layer: Adaptively determines "how many active units should be activated" (i.e., the proportion of active units) for each RIS, compressing the continuous high-dimensional problem into a small number of scalar parameters; Location layer: Given the constraint of the number of active nodes, the cells of the RIS node are scored and selected using Top-K based on the end-to-end link contribution, so that the active budget is preferentially allocated to the cell location that contributes the most to the end-to-end gain.

[0077] Through this decomposition, this step transforms the originally high-dimensional and difficult-to-solve unit-level optimization into a structured process of "proportion determination + sorting" that can be computed quickly, providing a low-latency and deployable RIS configuration mechanism for dynamic systems.

[0078] (1) Unit model of hybrid active / passive RIS nodes.

[0079] Let the r-th RIS node contain The configuration matrix of the units (active and passive) is represented as follows:

[0080] in Let n be the phase control quantity of the nth unit of the r-th RIS node. Amplitude coefficient Characterizing the type of the nth element: passive element takes Active unit ; This represents the amplitude gain of the active element. Using this unified model, the equivalent channel for any subsequent path can explicitly reflect the impact of the active element distribution on the end-to-end gain.

[0081] (2) Active ratio adaptive determination.

[0082] In multi-hop RIS node networks, different RIS nodes serve different "service distance scales": RIS nodes close to users or on short links can achieve high reflection efficiency even without passive operation; while RIS nodes at distant points or undertaking long-path cascading require more active gain compensation. If all RIS nodes use a fixed active ratio, it will result in wasted energy at the near end or insufficient gain at the far end. Therefore, this invention adaptively determines the proportion of active nodes based on the service distance index of the RIS nodes.

[0083] Let the set of service users of the r-th RIS node be . The service distance metric can be defined as the average distance:

[0084] in, Let be the distance from the r-th RIS node to user k.

[0085] This invention uses a smooth interpolation function to map the distance to the proportion of active cells in the r-th RIS node. :

[0086] in, , These are the pre-set near-distance threshold and far-distance threshold, respectively; , The pre-set ratios of near-end active nodes and far-end active nodes are determined based on factors such as RIS hardware capabilities, typical deployment distance range, and power consumption budget, and do not change dynamically with the environment.

[0087] Therefore, the number of active units that need to be activated in the r-th RIS node is:

[0088] in This indicates rounding down to the nearest integer.

[0089] The physical meaning of the above mechanism is: when RIS is in the proximal end ( When the RIS is small, the tendency is to reduce active cells to lower power consumption; when the RIS is at a remote location ( When the distance is large, the active power ratio should be appropriately increased to compensate for path loss. Furthermore, the ratio should change continuously with distance to avoid configuration jitter caused by hard handover.

[0090] (3) Unit contribution score and Top-K selection of RIS nodes.

[0091] Only the number of active units is determined. This is still insufficient to guarantee performance improvement. The key is to place active units in the "most effective" position within the RIS node unit array. Traditional methods of random selection or scoring based solely on a single link (e.g., only considering RIS and UE) easily overlook multi-hop cascade structures, making the activated units not necessarily the key contributors to end-to-end link gain.

[0092] Therefore, this invention proposes a unit contribution scoring function based on a multi-hop path structure: For unit n of the r-th RIS node, consider both its contribution from the previous hop link and the next hop link in the service path; let the nodes adjacent to the r-th RIS node in the path (which may be base stations, users, or other RIS nodes) be respectively... and ; In the currently selected service path set (i.e., all paths in which the r-th RIS node participates in the service) (a set) Above, define contribution metrics:

[0093] in , For the node Large-scale fading coefficient, incident channel component, or equivalent energy term up to the r-th RIS node; , For the r-th RIS node to node The large-scale fading coefficient, outgoing channel component, or equivalent energy term are used. The link fading coefficient can be obtained by constructing a path loss model or based on measured / empirical data. The incoming and outgoing channel components represent the channel components related to the unit n of the r-th RIS node, containing small-scale fading. They are usually complex numbers, containing both amplitude and phase information. Specifically, they can be obtained through statistical fading models, such as Rayleigh fading, or through measured / simulated data.

[0094] The key idea embodied in this contribution index function is that if a certain unit n is in a high-gain position in both the "incident to RIS" and "out of RIS" links, its contribution to the end-to-end cascade gain will be multiplicatively amplified, making it more worthy of being assigned as an active unit. After scoring is completed, this invention uses a greedy Top-K strategy to determine the set of active cells:

[0095] in, This indicates the selection of the indices of the K active units with the largest contribution indicators; and the amplitude coefficients are... The value assigned is:

[0096] In this way, the magnitude configuration of RIS changes from "continuous high-dimensional optimization" to "Top-K discrete selection," preserving structural gain while significantly reducing online complexity. Simultaneously, because... Depends on path structure and This mechanism is naturally coupled with the candidate paths in step 1 and the MIS grouping in step 2.

[0097] It should be noted that this step addresses the structured configuration problem on the RIS side regarding "how many active cells to activate and which cells to activate," but it does not yet provide the phase information. The final beamforming method also failed to jointly optimize the updated hybrid RIS configuration by mapping it to the end-to-end achievable rate. To transform the active set and amplitude configuration obtained in this step into system throughput improvement, it is necessary to further construct a multi-hop cascaded equivalent channel and perform joint precoding / beamforming and inter-group time slot allocation within the concurrent group, thereby forming a complete closed loop of "structure grouping - RIS configuration - physical layer joint optimization". This is the core problem that subsequent implementation steps will address.

[0098] Implementation Step 4: After completing candidate path selection, collision graph MIS grouping, and determining the proportion and location of hybrid active units, the system still needs to solve a key problem: how to truly translate the "path / grouping decision at the structural layer" and the "amplitude configuration at the RIS side" into an improvement in user rate. This is because the design of RIS phase shaping and base station precoding directly determines whether multi-hop cascaded signals can achieve coherent superposition at the user end; and within concurrent groups, residual mutual interference still exists between multiple users, requiring joint precoding and resource allocation to ensure fairness and stability. Without this physical layer closed loop, even if the greedy selection of active units is completed, the performance improvement may not be significant due to phase misalignment or excessive intra-group interference.

[0099] To this end, this invention proposes an integrated method of "equivalent channel-phase shaping-joint precoding-time slot allocation" in this step: first, an end-to-end cascaded equivalent channel is constructed based on the selected path, and then RIS phase conjugate shaping is performed using the hop-before-hop structure of the path to achieve coherent gain of the cascaded link; subsequently, linear precoding design is performed based on the equivalent channel in each concurrent group, and time slot ratio allocation is adopted between groups to maximize the system-level rate under the objective of multi-user fairness, thereby constructing a complete closed loop of "structural decision, physical layer optimization, and communication performance".

[0100] (1) Construction of multi-hop cascaded equivalent channels.

[0101] Let user k select service path p at the current time, with a node sequence of "base station, several RIS nodes, user". Since concurrently active path groups have structurally avoided conflicts of shared RIS nodes within the same group, the set of RIS nodes used by each path within the same concurrently active path group is usually non-conflicting, thus allowing phase shaping and equivalence processing to be performed independently on each path.

[0102] The end-to-end equivalent channel vector (from BS to user) for user k on path p can be expressed as follows:

[0103] In this context, the superscripts H and T in the parameters represent the conjugate transpose and the transpose, respectively. This represents the channel matrix from the base station to the first RIS node r1; Configuration matrix for RIS node r1; This is the channel matrix from the first RIS node r1 to the second RIS node r2; the other parameters have similar meanings. This is the configuration matrix for the Lth RISC node. Let L be the channel vector from the Lth RIS node to user k; there are a total of L RIS nodes.

[0104] Configuration matrix for each RIS node The amplitude coefficient has been determined in the previous steps. This step further provides the phase based on this. The shaping strategy.

[0105] (2) RIS phase conjugate shaping (using the “previous hop × next hop” structure to achieve cascaded coherent superposition).

[0106] In multi-hop cascading, the contribution of a RIS unit to the end-to-end gain comes from the joint phase of two links: the signal incident to the RIS and the signal emitted from the RIS to the next hop. If the phases are misaligned, the superposition of different units at the user end will cause cancellation, resulting in a severe reduction in cascade gain. To address this, this invention utilizes the path structure to construct the "local cascade coefficient" of each RIS unit and uses its phase conjugate as the shaping target, thereby obtaining low-latency coherent gain without high-dimensional iterative optimization.

[0107] For a certain RIS node on path p The local cascade complex coefficients of its nth unit are as follows:

[0108] in and These represent the previous and next hop nodes of the RIS in the path, respectively; They represent To RIS node RIS node arrive The incident channel component and the outgoing channel component.

[0109] The phase control of this unit adopts the conjugate alignment principle:

[0110] in, Represents RIS node The phase value of the nth unit on path p; This indicates taking the phase angle.

[0111] Therefore, the configuration matrix of RIS node r The nth diagonal element It can be written as:

[0112] in This is the final configuration of the nth unit of RIS node r; j It is the imaginary unit.

[0113] The significance of this phase shaping lies in the fact that, for each RIS unit, the conjugate of its "joint phase before and after hops" is used to compensate for the phase rotation of the cascaded links, so that the reflected (and amplified) signals from multiple units are superimposed in phase as much as possible at the next hop / user end, achieving high coherence gain with low complexity. Furthermore, since step 2 ensures that RIS units within the same group do not conflict, this shaping can be performed independently within the group, thereby meeting the low latency configuration requirements of dynamic scenarios.

[0114] (3) Intragroup joint precoding and rate expression.

[0115] Within the g-th concurrent activation path group, let the set of service users within the group be . The base station's downlink transmit signal vector for this group Using linear superposition:

[0116] in For user k, the precoded vector For corresponding data symbols; It indicates a desire for the expected value.

[0117] The user receives the following signal:

[0118] in, Let be the end-to-end equivalent channel vector for user k on path p. Added complex white Gaussian noise to the receiver of user k. It follows a Gaussian distribution. This represents noise power.

[0119] The user's intra-group signal-to-interference-plus-noise ratio (SINR) is:

[0120] in, j For service user set Another user in the group.

[0121] The corresponding intra-group instantaneous spectral efficiency is:

[0122] Intragroup precoding satisfies power constraints:

[0123] in, This represents the upper limit of the total transmission power of the base station. expressl 2-norm.

[0124] (4) Allocation of time slot ratio between groups.

[0125] Since step 2 yields multiple concurrently activated path groups, time-sharing scheduling is required between these groups. Let the time slot ratio occupied by the g-th concurrently activated path group be... Then the following condition is met:

[0126] Where G is the total number of concurrently activated path groups.

[0127] The average reachability of user k is:

[0128] in This represents the instantaneous spectral efficiency within the group.

[0129] To ensure fairness and dynamic stability among multiple users, this invention adopts a system-level objective of maximizing the average reachable rate of the minimum user (which can also be replaced by weighted fairness, proportional fairness, etc. in the implementation), and constructs a group-level equivalent channel model as follows:

[0130] Joint optimization of the group-level equivalent channel model yields the precoding vector. and time slot ratio .

[0131] Through the aforementioned closed loop: steps 1 / 2 / 3 provide path, grouping, and amplitude configuration; this step further provides phase shaping, precoding design, and time slot allocation, resulting in improved final performance. (Minimum rate, average rate, or total throughput, etc.) can be directly calculated and optimized, thereby realizing a verifiable link of "sensory architecture-communication performance".

[0132] Although this step constructs an equivalent channel and completes phase shaping, precoding, and time slot allocation, the channel and geometric relationships change over time in mobile scenarios. Executing steps 1–4 completely at every moment would incur high online overhead; reusing old solutions for extended periods would lead to performance degradation due to channel drift. Therefore, a dynamic low-latency update mechanism of "periodic re-optimization + rapid reuse evaluation at intermediate moments" must be designed to achieve stable, deployable online performance tracking. This is the core issue that step 5 will address in the subsequent implementation.

[0133] Implementation Step 5: In mobile communication scenarios, user location and environmental scattering change over time, causing the cascaded equivalent channel of the "base station-multiple RIS-user" link to exhibit significant time-varying characteristics. If steps 1 to 4 (candidate path generation, collision graph MIS grouping, hybrid active ratio and cell greedy algorithm, equivalent channel + phase shaping + precoding + time slot allocation) are fully executed at every moment, the online computational overhead will increase significantly, making it difficult to meet the requirements for low-latency real-time operation. Conversely, if old paths, groups, and RIS configurations are reused for a long time, beam alignment failure and interference control deteriorate due to channel drift, leading to a rapid decline in system speed and fairness. Therefore, this invention proposes a low-latency update mechanism for dynamic scenarios: "Periodic re-optimization" ensures structural adaptability, "intermediate-step fast reuse evaluation" reduces average computational load, and "trigger-time update" avoids significant performance degradation, thereby achieving a stable balance between computational overhead and dynamic performance.

[0134] (1) Time-varying channel and geometric update model (establishing a computable description of the "source of change").

[0135] To characterize link changes caused by movement at the system level, this invention establishes a time-dependent model for the time-varying small-scale components of any link. Let the link at time t... Small-scale channels are Then, a first-order autoregressive (AR(1)) model is used to describe its temporal evolution:

[0136] in The time correlation coefficient reflects the rate of change in user speed caused by the Doppler effect; This is an independent perturbation term. If we consider the large-scale loss varying with geometric distance, it can be written as:

[0137] Through the above model, this invention can uniformly describe the impact of "geometric changes + small-scale drift" on subsequent equivalent channel and rate indicators in dynamic simulation, providing a basis for triggering the update mechanism and performance closed-loop evaluation.

[0138] (2) Periodic re-optimization (to ensure structural tracking capability with a fixed update cycle).

[0139] Let the system's specified structural re-optimization cycle be... At time t, the following conditions are met:

[0140] The system performs a "full-process re-optimization update", including: Regenerate the candidate path set for each user based on the current geometry and link parameters (corresponding to step 1); Reconstruct the conflict graph and solve the MIS to form concurrent groups (corresponding to step 1); Based on the updated path structure and service distance metrics, recalculate the active ratio and Top-K active unit set for each RIS (corresponding to steps 2 and 3); Construct the updated equivalent channel and perform RIS phase conjugate shaping, intra-group precoding, and inter-group time slot allocation (corresponding to step 4).

[0141] The purpose of this periodic update is to "refresh structural decisions" at a controllable frequency, ensuring that the system does not gradually become mismatched due to long-term reuse of old structures, thereby maintaining robustness and fairness in dynamic scenarios.

[0142] (3) Intermediate step fast multiplexing evaluation (multiplexing structure and control variables, only update the equivalent channel and quickly calculate the rate).

[0143] At the non-re-optimization time, i.e., when t is satisfied:

[0144] This invention does not repeat the high-overhead structural calculations of steps 1 to 3, but instead reuses the most recent re-optimization time. What was obtained: Concurrent Grouping RIS amplitude configuration Phase shaping strategy Precoding vector Ratio of time slots Then, the equivalent channel for each user is reconstructed based solely on the current channel, and the SINR and rate within the group are calculated.

[0145] When precoding is reused, the SINR within a group is still calculated in real time based on the equivalent channel:

[0146] The instantaneous spectral efficiency within the group is:

[0147] And based on this, the long-term average rate (multiplexed time slot ratio) is obtained:

[0148] The key advantage of this "intermediate step rapid evaluation" is that it avoids repeatedly solving discrete structures (paths, MIS) and cell-level configurations (Top-K) at each step, allowing online computation to focus mainly on the lightweight step of "equivalent channel update + rate calculation", thereby significantly reducing average overhead while still reflecting the impact of channel drift on performance in real time.

[0149] (4) Triggered early update (when performance degradation exceeds the threshold, the fixed cycle is broken to avoid mismatch spread).

[0150] Relying solely on fixed-period updates can lead to a "rapid performance collapse within a period" problem during high-speed movement. Therefore, this invention introduces a triggered early update mechanism: when the minimum user rate or key fairness indicators are detected to have significantly decreased over a consecutive period of time, a full-process re-optimization is performed in advance to avoid the cumulative spread of beam mismatch and increased interference.

[0151] Define the minimum user rate of the system at time t as:

[0152] Let the time of the most recent re-optimization be... The corresponding minimum reference rate is :

[0153] One of the triggering conditions can be defined as a relative descent threshold. :

[0154] To avoid frequent updates caused by instantaneous fluctuations, this invention further adopts a "triggered by M consecutive fulfillments" strategy, defining an indicator function. :

[0155] Once triggered, the system immediately executes a full-process update from steps 1 to 4, refreshing the structure and control variables to a solution that better matches the current channel, thereby ensuring that performance stability and fairness do not collapse in dynamic scenarios.

[0156] In summary, this invention, through parameterized candidate path generation in step 1, concurrent grouping of conflict graph MIS in step 2, active proportional adaptive and Top-K unit selection in step 3, equivalent channel and phase shaping / precoding / time slot allocation in step 4, and the introduction of a dynamic low-latency update mechanism in step 5, ultimately forms a deployable integrated closed loop of "structure-physical layer-performance monitoring". This enables the system to maintain stable throughput and fairness in mobile scenarios, while meeting the constraints of low-latency online operation.

[0157] In the method proposed in this invention, once the generation of multi-hop candidate paths and the determination of concurrent feasibility of user services are completed through the aforementioned steps, the system will immediately execute structured collaborative configuration and physical layer closed-loop optimization for multi-RIS multi-hop links.

[0158] 1. First, the system selects candidate paths from each user's path set based on path cost and received power threshold. The optimal service path at the current moment is obtained through filtering. And construct a path conflict graph to form a set of groups that can be activated concurrently. This process makes the "multi-user interference coupling problem" explicit and controllable at the structural layer, laying the foundation for subsequent stable concurrent transmission.

[0159] 2. Next, for each scheduled RIS node r, the system adaptively determines the active ratio and obtains the number of active units according to the service distance scale. Then, the RIS units are selected using a Top-K selection based on the joint contribution score of the "previous hop × next hop" to obtain the active set. Based on this, the RIS controller combines amplitude configuration with phase conjugate alignment principles to calculate and distribute the corresponding RIS control matrix. This allows for the coherent superposition of multi-hop cascaded links at the user end, thereby prioritizing the allocation of limited active budgets to the most gain-sensitive end-to-end locations.

[0160] 3. Furthermore, to maximize communication performance within the concurrent group, the system constructs an end-to-end equivalent channel based on the selected path and RIS configuration. And linear precoding is used within the group. Suppress residual multi-user interference, while using time slot ratios between groups. Time-sharing scheduling is employed to balance fairness and throughput. This process does not require high-dimensional iterative joint optimization of all RIS units; instead, it relies on a strategy of "structural decomposition + low-dimensional parameters + greedy selection" to significantly reduce online computation latency while ensuring performance.

[0161] Once the above configuration is completed, in order to verify the final effectiveness of the overall structure of the present invention in dynamic mobile scenarios, the present invention constructs a performance verification closed loop of "structural decision-making, physical layer control, and communication performance".

[0162] 1. First, given the grouping, path, and RIS configuration, the system calculates the reachable rate for each user. and minimum rate index This will be used to measure the actual benefits of concurrent scheduling versus active allocation.

[0163] 2. Then, compare the result with the ideal performance upper limit, for example, the theoretical upper limit under the condition of having a globally optimal structure / perfect equivalent channel, or the reference upper limit obtained by using full activation / global joint optimization (not achievable in real time), so as to characterize the performance gap of the "structured low latency scheme" relative to the ideal upper limit.

[0164] 3. Finally, the system introduces a dynamic low-latency update mechanism: Under the joint constraints of a fixed update cycle and a trigger-based performance threshold, the path, group, and RIS active distribution are periodically re-optimized and intermediate steps are quickly reused and evaluated to avoid the cumulative spread of structural mismatch caused by movement and to ensure the continuous high quality and fairness of the link in dynamic scenarios.

[0165] Thus, the method of this invention not only proposes a holistic structured framework for multi-RIS multi-hop mobile networks, but also tightly couples structural layer decisions with physical layer communication performance through an integrated design of "candidate path - conflict graph grouping - active adaptive greedy algorithm - equivalent channel precoding - dynamic update closed loop". Finally, its practical value in dynamic scenarios is demonstrated through system-level closed-loop evaluation. This process ensures that the proposed low-latency structured RIS collaborative method can seamlessly support high-reliability, high-throughput mobile communication, providing solid theoretical and engineering support for RIS to move from static deployment to real-time dynamic network applications.

[0166] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A low-latency adaptive configuration method for multi-hop RIS-assisted communication, characterized in that, include: In a network consisting of a base station, multiple active / passive RIS nodes, and multiple users, a candidate path set containing at least one RIS node is generated for each user. During the generation process, maximum hop count and received power threshold constraints are introduced to control complexity. A path conflict graph is constructed, and the shared RIS nodes or path pairs with resource contention between multiple user paths are explicitly modeled as conflict edges in a graph structure. By solving for the maximum independent set or the equivalent set of conflict-free path groups, concurrently activated path groups are obtained. For each hybrid active / passive RIS node, based on the user distance information corresponding to the path served by the RIS node, a smooth interpolation mechanism is used between the near distance threshold and the far distance threshold to adaptively determine the proportion of active units in each RIS node. After determining the proportion of active units, a contribution index is constructed for all units within each hybrid active / passive RIS node. This contribution index reflects the joint gain of the unit in the two adjacent links in the selected path, and is weighted and accumulated by combining the large-scale fading coefficients of the two links. Then, units are sorted according to the contribution index, and a preset number of units are selected as the active unit set. The group-activated multi-hop links are mapped to a group-level equivalent channel model, thereby converting the combined effect of multi-hop links on users into an equivalent expression that can be used for joint optimization. On this group-level equivalent channel model, the allocation optimization of joint precoding vectors and time slot ratios is performed to maximize the minimum average reachability of the system.

2. The low-latency adaptive configuration method for multi-hop RIS-assisted communication according to claim 1, characterized in that, For each user, a candidate path set containing at least one RIS node is generated, and maximum hop count and received power threshold constraints are introduced during the generation process to control complexity, including: The path generation problem for each user is transformed into a constrained shortest path search problem: for each user... k On the node graph, from the base station node To user node Search candidate path set And apply maximum hop count and receive power threshold constraints: ； ； in, Indicates selection The smallest front K Path; Representing paths The starting node and any node; Indicates the path direction; For path The number of jumps; The maximum number of hops allowed; This is the reachability threshold for the received power; For path The cumulative cost of loss; To estimate the received power: ； in, This refers to the base station's transmission power. For path Path loss, This represents the power gain of a single GIS node.

3. The low-latency adaptive configuration method for multi-hop RIS-assisted communication according to claim 1, characterized in that, The process of constructing a path conflict graph is as follows: Let the set of RIS nodes used by any path be denoted as . Two paths are considered to be in conflict if either of them satisfies one of the following conditions: Shared RIS nodes; RIS node clusters are spatially adjacent; Therefore, a conflict determination function is constructed. Conflict determination function A value of 1 indicates a path Conflict, 0 indicates no conflict; Construct a path conflict graph ;in Let the set of edges be defined as follows: if the paths between two vertices are... Conflict conditions are met If two paths cannot be activated simultaneously, then an edge is drawn between them; vertex set It consists of "user-path" pairs, where each vertex represents a candidate service option. It can be written as: ； in, For user collection; For users k The set of candidate paths; if the two paths corresponding to two vertices conflict, then connect them with an edge.

4. The low-latency adaptive configuration method for multi-hop RIS-assisted communication according to claim 1, characterized in that, By solving for the maximum independent set or the equivalent set of conflict-free path groups, concurrently active path groups are obtained, including: In a path conflict graph, a concurrently activated path group must satisfy the following condition: any two vertices within the group... There are no conflicting edges between them; let the first... g The concurrent activation path groups are denoted as follows: Then the independence constraint is: ； in, Represents two vertices There is no edge between them; Concurrent activation path grouping is constructed using the maximum independent set criterion: ； in, Represents the vertex set A subset of; This represents a function that finds the index of the maximum value.

5. The low-latency adaptive configuration method for multi-hop RIS-assisted communication according to claim 1, characterized in that, A smooth interpolation mechanism is used between the near-distance threshold and the far-distance threshold to adaptively determine the proportion of active units in each RIS node, including: Let the set of service users of the r-th RIS node be . The service distance metric is defined as the average distance: ； in, Let be the distance from the r-th RIS node to user k; A smooth interpolation function is used to map the distance to the proportion of active cells in the r-th RIS node. : ； in, , These are the pre-set near-distance threshold and far-distance threshold, respectively; , The pre-set ratio reference values ​​for near-end active nodes and far-end active nodes; Therefore, the number of active units that need to be activated in the r-th RIS node is: ； in This indicates rounding down to the nearest integer.

6. The low-latency adaptive configuration method for multi-hop RIS-assisted communication according to claim 1, characterized in that, Contribution metrics are constructed for all units within each hybrid active / passive RIS node; a preset number of units are selected as the active unit set, including: For cell n of the r-th RIS node, let the preceding and following nodes adjacent to the r-th RIS node in the path be respectively... and In the currently selected service path set Above, define the contribution index: ； in , For the node Large-scale fading coefficient and incident channel component up to the r-th RIS node; , For the r-th RIS node to node Large-scale fading coefficient and outgoing channel components; The set of active cells is determined using a greedy Top-K strategy: ； in, This indicates the selection of the indices of the K active units with the largest contribution indicators; and the amplitude coefficients are... The value assigned is: ； in, This represents the amplitude gain of the active unit.

7. The low-latency adaptive configuration method for multi-hop RIS-assisted communication according to claim 1, characterized in that, For a certain RIS node on path p The local cascade complex coefficients of its nth unit are as follows: ； in and These represent the previous and next hop nodes of the RIS in the path, respectively; They represent To RIS node RIS node arrive The incident channel component and the outgoing channel component; The phase control of this unit adopts the conjugate alignment principle: ； in, Represents RIS node The phase value of the nth unit on path p; This indicates taking the phase angle; Therefore, the configuration matrix of RIS node r The nth diagonal element It can be written as: ； in This is the final configuration of the nth unit of RIS node r; j It is the imaginary unit.

8. The low-latency adaptive configuration method for multi-hop RIS-assisted communication according to claim 1, characterized in that, The group-level equivalent channel model is constructed as follows: ； Joint optimization of the group-level equivalent channel model yields the precoding vector. and time slot ratio ; This represents the maximum total transmit power of the base station. express l 2-norm; Let the time slot ratio occupied by the g-th concurrent activation path group be... Then the following condition is met: ； Where G is the total number of concurrently activated path groups; The average reachability of user k is: ； Let g be the set of service users within the g-th concurrent activation path group; Intra-group instantaneous spectral efficiency: ； in Let S be the signal-to-interference-plus-noise ratio (SIR) for user k within the concurrent active path group.

9. A terminal device, comprising a processor, a memory, and a computer program stored in the memory; characterized in that, When the processor executes a computer program, it implements the low-latency adaptive configuration method for multi-hop RIS-assisted communication as described in any one of claims 1-8.

10. A computer-readable storage medium storing a computer program; characterized in that, When the computer program is executed by the processor, it implements the low-latency adaptive configuration method for multi-hop RIS-assisted communication as described in any one of claims 1-8.

Images