CN-121357160-B - Heterogeneous resource deterministic arrangement and distribution method in holographic communication intelligent identification network

Abstract

The invention provides a deterministic arrangement and distribution method for heterogeneous resources in a holographic communication intelligent identification network. The method comprises the steps that edge computing equipment at a transmitting end in a DRHA system in a holographic communication intelligent identification network processes PCF, the PCF is divided into a core structure part PCF and a surface detail part PCF through CTS, a supported HQF switch utilizes an MRSA algorithm to calculate, store and coordinate network resources, and the core structure part PCF and the surface detail part PCF are transmitted through different queues and time slots. And after decompression processing is carried out on the received core structure part PCF and the surface detail part PCF by edge computing equipment of a receiving end in the DRHA system, the reconstructed point cloud is rendered on a display. The method and the device can effectively relieve the dynamic fluctuation influence of the point cloud frame, improve the transmission quality and user experience of the holographic video stream, and ensure bounded ultra-low delay transmission and deterministic resource allocation.

Inventors

• CHEN JIA
• DAI YAJIE
• ZHENG DEKAI
• HUANG XU
• Man Jiaqian
• QIAN DONGSHENG
• XUE JING
• Liao Chenqian
• LI WEICHAO
• LIU JINGWU
• DONG KAIYING

Assignees

• 北京交通大学

Dates

Publication Date

20260512

Application Date

20251104

Claims (3)

1. 1. A heterogeneous resource deterministic arrangement and distribution method in a holographic communication intelligent fusion identification network is characterized by comprising the following steps: An edge computing device of a transmitting end in a heterogeneous resource distribution system DRHA in a holographic communication intelligent identification network processes a point cloud frame PCF generated by a user end, and the PCF is divided into a core structure part PCF and a surface detail part PCF through a point cloud shaper CTS; The supported switch for forwarding the HQF in a layered queuing way utilizes a multi-user resource scheduling algorithm MRSA algorithm to calculate, store and coordinate network resources of the PCF of the core structure part and the PCF of the surface detail part, and transmits the PCF of the core structure part and the PCF of the surface detail part through different queues and time slots; After decompression processing is carried out on the received core structure part PCF and the surface detail part PCF, the edge computing equipment of the receiving end in the DRHA system renders the reconstructed point cloud on a display of each user end; The supported HQF switch performs computation, storage and coordinated allocation of network resources of the core PCF and the surface detail PCF by using an MRSA algorithm, and transmits the core PCF and the surface detail PCF through different queues and time slots, including: The supported HQF exchanger compresses the core structure part into a single file, divides the surface detail part into a plurality of maximum transmission unit MTU-level data packets, and compresses each MTU-level data packet into a single binary file respectively; in the upper layer queue of the switch, PCF is used as a dispatching basic unit to transmit various PCFs with different priorities, and in the lower layer queue of the switch, a single MTU-level data packet is used as a dispatching basic unit to transmit each MTU-level data packet according to different orders and rates; Setting constraint conditions of the switch for transmitting MTU-level data packets and core structure part files through a deep Q network DQN (digital to analog) comprises calculation delay constraint, storage and network delay constraint, MTP (modulation transfer protocol) delay constraint and heterogeneous resource transaction constraint, and setting optimization problem based on the constraint conditions, the compression ratio alpha and the optimal value of a sampling factor gamma so as to maximize DRHA system capacity as a target while meeting bounded MTP delay constraint, wherein the optimization problem is expressed as: (10) Wherein the method comprises the steps of The DRHA system is shown as accommodating the maximum number of users, Representing the total number of potential users, Representing the upper bound MTP delay required for DRHA systems, As a set of compression ratios, As a set of sampling factors, Indicating MTP delay for any one user, the upper limit is , Representing the number of users that can be served under the current action set, For all user numbers; Solving the optimization problem through DQN to any user Assigning corresponding compression ratios And sampling factor According to the compression rate And sampling factor Calculate the user The required computing power resource and the network time slot requirement are obtained to obtain the user MTP delay of (C) When (when) Less than maximum MTP delay upper bound When determining the user It is possible to successfully schedule the scheduling of the data, Representing a current set of actions The number of users that can be successfully scheduled, For all the user sets, calculate the current action set Successful scheduling rate under And rewards According to rewards Adjusting a set of actions And (3) performing next round of DQN iterative training, re-distributing corresponding compression rate and sampling factor to each user, re-calculating success scheduling rate and rewarding until the difference between the success scheduling rate and a set threshold is smaller than a set numerical range or the set iterative training times are reached, and finishing the DQN iterative training to obtain the optimal computing power resource size and network time slot requirement distributed to each user, and transmitting the core structural part PCF and the surface detail part PCF of the user through different queues and time slots according to the computing power resource and the network time slot requirement of the user.
2. 2. The method of claim 1 wherein said dividing said PCF into a core fabric portion PCF and a surface detail portion PCF by a point cloud shaper CTS comprises: The number of core structure points in each PCF is set to be constant, and the definition is as follows: (1) where γ represents a sampling factor determined by the visual quality requirements of the user for the user Point cloud frame sequence Any one frame of (3) The point cloud data volume processed by the sampling factor gamma is that , For point cloud frames To the number of point clouds of (a) For users The maximum number of point clouds allowed to be transmitted; data size of PCF of core Structure part Calculated by: (2) For users The maximum number of point clouds allowed to be transmitted, 28Byte is the data volume of a single point cloud, For the data volume of all the point clouds, For the point cloud file header size, For users The maximum point cloud frame file size allowed to be transmitted; User' s Is the number of time slots of (a) The following equation gives: (3) Wherein the method comprises the steps of The bandwidth is represented by a bandwidth that, Representing the size of one cycle; After the PCS extracts the core structure portion PCF from the overall PCF, the remaining PCFs are classified as surface detail portion PCFs.
3. 3. The method of claim 1, wherein the setting constraint conditions for the switch to transmit MTU level packets and core fabric portion files over the deep Q network DQN includes calculating delay constraints, storage and network delay constraints, MTP delay constraints, and heterogeneous resource transaction constraints, comprising: Setting the exchange to transmit subscribers through DQN The delay constraint of the core structure portion file of (a) is: ; (4) Wherein the method comprises the steps of Representing through-switches Is a set of user flows of (a), Representing a user Is provided with a routing path for the routing path, For users Is provided with an end-to-end delay, For users The number of time slots to be occupied, Is a single slot size; Setting calculation delay constraint of the switch for transmitting MTU-level data packets through DQN as follows: (5) Where τ represents that the process corresponds to The time required for the point cloud data of the volume, For users The number of time slots required, representing the time required to process point cloud data within a time slot, For users The required processing calculates the time delay; User' s The corresponding compression and decompression times are expressed as: (6) Wherein the function is ( ) Indicating use Time required for each thread and compressing point cloud data with alpha being Size ( ) Then the time required to decompress the same data size using the same number of threads is represented; The computational delay for setting the switch to transmit the core fabric portion PCF data over the DQN is expressed as: (7) in order to handle the time delay, In order to compress the time delay, In order to decompress the time delay, For users Total computation delay; Setting an end-to-end transmission delay of the switch for transmitting the MTU level data packets through the DQN to be expressed as: (8) for the time delay of the network transmission, In order to store the time delay, Is an end-to-end delay; setting the total delay of the switch for transmitting MTU-level data packets via DQN The method comprises the following steps: (9) in order to handle the time delay, In order to compress the time delay, In order to decompress the time delay, Is an end-to-end delay.

Description

Heterogeneous resource deterministic arrangement and distribution method in holographic communication intelligent identification network Technical Field The invention relates to the technical field of mobile communication, in particular to a heterogeneous resource deterministic arrangement and distribution method in a holographic communication intelligent identification network. Background With the rise of sixth generation mobile communication technology (6G), metauniverse (METAVERSE) and CPN (Computing Power Network, power network), HTC (Holograghy Type Communication, holographic-type communication) is gradually becoming an important trend of future communication modes. HTC utilizes volumetric video technology to bring the user with an immersive experience that is immersive. Holographic communication has the characteristics of large bandwidth, low time delay and strong calculation power, and needs the network to have the capability of self-adaptive dispatching calculation, forwarding, storage and other diversified resources. The multi-user holographic conference is an emerging communication technology, and real-time 3D interaction and cooperation of a plurality of participants in a virtual space are realized by using point cloud video as a core carrier. The conference form surpasses the traditional video call, provides a highly immersive holographic projection experience, and enables users to walk freely, observe three-dimensional figures of each other and perform natural interaction through gestures or voices as if the users were in the same physical conference room. In a multi-user holographic conference, the holographic projection of each participant is constructed by PCFs (Point Cloud Forecasting, point cloud frames) containing thousands of points with precise coordinates and colors, which together render a dynamic 3D mannequin or environmental scene, supporting real-time perspective switching from any angle. In order to support smooth multi-person interaction, a typical conference scenario may involve 4-10 users, a holographic stream of 30 point cloud frames per Second FPS (FRAMES PER seconds, number of transmission point cloud frames per Second) needs to process multiple synchronized point cloud frames, the total data amount may vary from hundreds of MB to several GB, thus requiring transmission rates as high as 10-50Gbps, even higher, which poses serious challenges for 5G/6G networks and edge computing infrastructure, but also opens a new era of remote collaboration. In the deterministic transmission context of multi-user holographic conferences, the introduction of depth Q network-based multi-user holographic conference deterministic DHRA (DETERMINISTIC HETEROGENEOUS RESOURCE ALLOCATION, heterogeneous resource allocation system) provides powerful support for efficient resource management. DHRA through deep reinforcement learning and dynamic resource perception capability, heterogeneous resources such as calculation, network and storage can be intelligently allocated, a scheduling path of point cloud data processing is optimized, and transmission uncertainty of huge data volume and high fluctuation is relieved. A real-time volume video transmission method in the prior art comprises a technology of transmitting three-dimensional volume video data through a network, wherein the technology is used for transmitting complex three-dimensional space data in real time so as to conduct real-time rendering, display and interaction at a receiving end. A system for visually perceived volumetric video streaming media on a mobile device is presented that can save an average of 40% of the data usage while maintaining visual quality. The method realizes high-efficiency transmission and decoding of the high-fidelity volume video through an innovative PD-Tree (Parallel Decodable Tree ) data structure and a color information compression technology. The disadvantages of the prior art methods of real-time video transmission include that they focus on efficient data compression and partly on data streaming, but they often ignore how to ensure accurate control and consistency during transmission. In an uncertain network environment, fluctuations and loss of data may cause degradation of image quality and further increase delay. This instability makes the prior art difficult to use widely in practical holographic communication systems. Disclosure of Invention The embodiment of the invention provides a heterogeneous resource deterministic arrangement and distribution method in a holographic communication intelligent identification network, which is used for realizing low-delay and high-deterministic transmission end to end in a multi-user holographic conference scene. In order to achieve the above purpose, the present invention adopts the following technical scheme. A heterogeneous resource deterministic arrangement and distribution method in a holographic communication intelligent fusion identification network comprises th