CN-122027987-A - Low communication overhead position measurement and resource scheduling method

CN122027987ACN 122027987 ACN122027987 ACN 122027987ACN-122027987-A

Abstract

In order to remarkably reduce communication resource consumption in a wireless co-location system, the invention provides a low communication overhead position measurement and resource scheduling method, and relates to the technical field of wireless communication and location. The method comprises the steps of modeling a positioning network into a directed graph, dynamically evaluating link quality by utilizing a graph attention mechanism, calculating communication weight among nodes, realizing intelligent resource scheduling for a high-quality link, designing a shared scalar quantization codebook, compressing a high-dimensional node state vector into a low-bit codeword index for transmission, and greatly reducing communication data volume. Simulation and hardware prototype experiments show that the method can reduce the communication overhead by about 60% in the co-location process on the premise of keeping high location precision, and effectively solves the expansibility problem of high-precision location in a resource-limited wireless network.

Inventors

SHU YUQUAN
YE HONGJUN
XU LEI
SUN BO
ZHANG YALIN
MA ZHIQI
MU JUNSHENG

Assignees

中国电子科技集团公司第五十四研究所

Dates

Publication Date: 20260512
Application Date: 20260130

Claims (7)

1. A method of low communication overhead location measurement and resource scheduling for a wireless network comprising an anchor node and a proxy node, wherein the anchor node location is known and the proxy node location is unknown, the method comprising the steps of: The method comprises the steps of S1, modeling an anchor node, an agent node and a measuring link between the anchor node and the agent node, which are needed to participate in information exchange, in a wireless network as a graph model, wherein the anchor node takes self known position information as a core attribute to participate in the graph model construction, the agent node takes self initial position estimated information and link measuring data as the core attribute to participate in the graph model construction, and an effective measuring link is established as an edge of a graph when the physical distance between the nodes is not more than a preset communication radius; S2, based on a graph attention mechanism, the anchor node and the proxy node dynamically calculate the communication weight between each neighbor node according to the self state and combining the characteristics of the sending node, the characteristics of the receiving neighbor node and the characteristics of the link, and perform differentiated scheduling of the communication resources according to the communication weight to allocate more communication resources to the link with higher weight; s3, the nodes needing to participate in information exchange, including anchor nodes and proxy nodes, compress the state information of the nodes through a shared quantization codebook, generate corresponding compression identifiers and transmit the compression identifiers to neighbor nodes, and at a receiving end, the neighbor nodes recover the state information from the quantization codebook according to the compression identifiers; S4, the agent node iteratively calculates the self accurate position based on the recovered neighbor node state information.
2. The method for measuring the position and scheduling the resources with low communication overhead according to claim 1, wherein in S2, a multi-head attention mechanism is adopted to calculate a plurality of groups of different communication weights in parallel, each group of weights captures the characteristics of a transmitting node, the characteristics of a receiving neighbor node and the characteristics of a link respectively from different dimensions, a plurality of groups of information obtained by aggregation based on the communication weights of each group are fused to obtain node state update data, wherein the fusion result of an anchor node is the self-known position information and the position stability state data after strengthening, and the fusion result of an agent node is the position estimated information and the link quality comprehensive evaluation data after correcting.
3. The method for low communication overhead location measurement and resource scheduling according to claim 1, wherein the S3 concrete procedure comprises: Pre-training a quantized codebook comprising a plurality of codeword shares, each codeword corresponding to a representative value of a high-dimensional state vector; The nodes needing to participate in information exchange comprise an anchor node and an agent node, wherein the state information of the anchor node is reference data, comprises self known positions and communication states, and the state information transmitted by the agent node is dynamic data, comprises self initial position estimation information and link quality feedback; At the receiving end, the anchor node receives the dynamic data of the neighbor proxy node, the proxy node receives the reference data of the neighbor anchor node and the dynamic data of other proxy nodes, and the corresponding code word is searched from the quantization codebook according to the compression identification, so that the state information is recovered.
4. The low communication overhead location measurement and resource scheduling method of claim 3 wherein the amount of data of the compressed identity is substantially less than the amount of data of the original node state vector.
5. The method for measuring the position and scheduling the resources with low communication overhead according to claim 1, wherein the method is implemented on a heterogeneous hardware platform comprising a programmable logic unit and a processing system, wherein in S2, an anchor node and an agent node are respectively operated in the processing system according to self states based on a graph attention mechanism, combined with a sending node characteristic, a receiving neighbor node characteristic and a link characteristic, a processing process of dynamically calculating a communication weight between the anchor node and each neighbor node, and a processing process of compressing self state information through a shared quantization codebook in S3, wherein the processing process of compressing the self state information through the anchor node and the agent node in the S3 is implemented in the programmable logic unit, the transmission link process of generating a corresponding compression identifier and transmitting the corresponding compression identifier to the neighbor node and the cooperative control of the whole flows of S1 to S4 are implemented in the processing system, and the programmable logic unit and the processing system perform high-speed data interaction through a direct memory access mode.
6. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the method of any one of claims 1 to 5 when executing the computer program.
7. A computer readable storage medium, on which a computer program is stored, characterized in that the program, when being executed by a processor, implements the method according to any one of claims 1 to 5.

Description

Low communication overhead position measurement and resource scheduling method Technical Field The invention relates to the technical field of wireless communication and positioning, in particular to a method for high-precision cooperative positioning in a wireless network with limited resources, and particularly relates to a low-communication overhead position measurement and resource scheduling method based on a graph attention neural network and a state quantization technology. Background With the rapid development of the internet of things, industrial automation and unmanned systems (such as unmanned aerial vehicle clusters and automatic driving motorcades), the demand for high-precision and real-time co-location of mobile targets in wireless networks and network coverage is increasingly urgent. The co-location technology realizes the position calculation of the nodes with unknown positions through the mutual distance measurement and information exchange among the nodes, and is one of key technologies for meeting the requirements. Conventional co-location methods, such as iterative algorithms based on least squares or distributed algorithms based on kalman filtering, typically rely on frequent, complete exchange of position state vectors or raw measurement data between nodes. When the network scale is enlarged or the node density is increased, the communication mode can generate huge wireless channel overhead, so that network congestion and positioning delay are increased, and the expandability and the instantaneity of the system are seriously restricted. Particularly in the context of bandwidth-limited industrial internet of things or wide area network protocols with low power consumption, communication resources have become a major bottleneck in improving positioning performance. In order to reduce communication overhead, the prior art mainly optimizes from two directions, namely, communication topology optimization, such as designing sparse information exchange rules based on a graph model, but the method usually adopts static or heuristic rules and cannot carry out self-adaptive adjustment according to dynamically-changed channel quality and positioning requirements, and data compression, such as traditional lossy compression coding, is adopted, but the general compression method does not consider data characteristics and error propagation characteristics of positioning tasks, and is difficult to achieve good balance between compression rate and positioning accuracy. In addition, the actual wireless environment is complex and changeable, the link quality difference is obvious and the possibility of burst interruption exists. The prior method generally lacks a robust processing mechanism for the dynamic factors, and the performance of the method is obviously reduced in a complex environment. Therefore, a novel co-location method capable of intelligently sensing network status, dynamically scheduling communication resources and greatly reducing communication overhead on the premise of ensuring location accuracy is needed. Disclosure of Invention In order to overcome the defects of the prior art, the invention provides a low-communication-overhead position measurement and resource scheduling method. The invention has the core purpose of obviously reducing the communication data volume between nodes while ensuring the co-location precision through intelligent information screening and high-efficiency data compression. In order to achieve the above purpose, the invention adopts the following technical scheme: A method of low communication overhead location measurement and resource scheduling for a wireless network comprising anchor nodes and proxy nodes, wherein the anchor node locations are known and the proxy node locations are unknown, the method comprising the steps of: The method comprises the steps of S1, modeling an anchor node, an agent node and a measuring link between the anchor node and the agent node, which are needed to participate in information exchange, in a wireless network as a graph model, wherein the anchor node takes self known position information as a core attribute to participate in the graph model construction, the agent node takes self initial position estimated information and link measuring data as the core attribute to participate in the graph model construction, and an effective measuring link is established as an edge of a graph when the physical distance between the nodes is not more than a preset communication radius; S2, based on a graph attention mechanism, the anchor node and the proxy node dynamically calculate the communication weight between each neighbor node according to the self state and combining the characteristics of the sending node, the characteristics of the receiving neighbor node and the characteristics of the link, and perform differentiated scheduling of the communication resources according to the communication weight to allocate more communication re