System and method for dynamic redundancy-aware blockchain-based partial computation offloading for metaverse within computing environment in network

Abstract

The present disclosure relates to a system and method for dynamic redundancy-aware blockchain-based partial computation offloading for a metaverse within a computing environment in a network, and according to the present disclosure, in order to improve the QoS of a metaverse service, it is possible to provide an environment that may perform existing traditional task offloading, perform optimal offloading through an in-network computing agent, and provide an expandable network and an ultra-low latency service. In addition, it is possible to maximize incentives while minimizing the overhead of computation execution costs incurred when a user performs a task, and satisfy the constraints on latency and blockchain offloading costs.

Description

A dynamic redundancy-aware blockchain-based partial computation offloading system and method for the metaverse within an in-network computing environment.

[0001] The present disclosure relates to a method and system for optimizing offloading for applications and services in an in-network computing environment.

[0002] Metaverse-based services provide users with immersive experiences based on technologies such as virtual reality (VR) and augmented reality (AR).

[0003] However, these metaverse-based services have various limitations when applying user equipment (UE) or mobile edge computing (MEC).

[0004] Network optimization techniques for allocating network resources for metaverse services are crucial for network performance that satisfies QoS (Quality of Service) while simultaneously ensuring user security and operating metaverse services.

[0005] In contrast, the paradigm of in-network computing (COIN) utilizes even underutilized network resources to minimize network latency and optimize user immersive experiences. However, as the use of in-network computing increases, power consumption increases nonlinearly, making optimization essential for building metaverse services.

[0006] Blockchain provides privacy and secure communication for distributed data, but it also poses scalability issues, such as high storage demands due to complete data duplication, which impacts data decentralization.

[0007] For example, Bitcoin's blockchain exceeds 477GB and requires 10,000 nodes for approximately 4.6PG of storage to be secured.

[0008] Therefore, resolving data redundancy in a network environment involves a complex optimization problem that takes into account the cost of the blockchain, mining incentives, and the cost of decentralization and offloading, and requires efficient computation offloading to calculate this.

[0009] The present disclosure was conceived to solve the above problems, and its primary task is to provide a dynamic redundancy-aware blockchain-based partial computation offloading system and method for a metaverse within an in-network computing environment.

[0010] A dynamic redundancy-aware blockchain-based partial computation offloading system for a metaverse within an in-network computing environment according to at least one of various embodiments of the present disclosure comprises: a user device transmitting a request for a metaverse task; And an in-network computing device that operates a metaverse service, wherein the in-network computing device includes a plurality of layers and a controller that calculates partial operation offloading and transmits a request to determine a layer to perform a metaverse task requested by the user equipment, wherein the plurality of layers includes a first layer that receives user data including a task request transmitted from the user equipment and performs first task offloading or transmits it to a metaverse service network end, wherein the plurality of layers includes a second layer that receives user data containing the task request through the first layer and performs the first task offloading or partial operation offloading, and wherein the plurality of layers includes an offloading policy for determining partial operation offloading and a third layer that ensures stability and privacy of data containing the user request so as to be able to recognize whether redundancy is recognized, wherein the controller calculates a task offloading policy based on a current network status when a task is requested from a user terminal, and if the calculation is performed locally, it is executed in a corresponding layer, and if the calculation is not performed locally, it determines an offloading destination based on a cost and a queue status as a first offloading destination. It can be configured to determine one of the first layer, the second layer, and the third layer.

[0011] In addition, a dynamic redundancy-aware blockchain-based partial computation offloading method for a metaverse in an in-network computing environment according to at least one of various embodiments of the present disclosure comprises the steps of: receiving a request for a metaverse task from a user device; calculating partial computation offloading through a plurality of layers; And a step of determining a layer to perform a metaverse task requested by the user equipment by the controller, wherein the plurality of layers includes a first layer that receives user data including a task request transmitted from the user equipment and performs first task offloading or transmits the task to a metaverse service network end, the plurality of layers includes a second layer that receives user data containing the task request through the first layer and performs the first task offloading or partial computation offloading, and the plurality of layers includes a third layer that ensures stability and privacy of data containing the user request so as to know whether an offloading policy for determining partial computation offloading and whether redundancy is recognized, and the controller may be configured to calculate a task offloading policy based on a current network status when a task is requested from a user terminal, and to execute the task offloading policy in a corresponding layer when the calculation is performed locally, and to determine an offloading destination as one of the first layer, the second layer, and the third layer based on a cost and a queue status when the calculation is not performed locally.

[0012] According to an embodiment of the present disclosure, the following effects are achieved.

[0013] First, in order to improve the QoS of metaverse services, it not only performs traditional task offloading, but also performs optimal offloading through in-network computing agents, and has the effect of providing an environment that can provide scalable networks and ultra-low latency services.

[0014] Second, it has the effect of maximizing incentives while minimizing the overhead of computational execution costs incurred when users perform tasks, and satisfying constraints on delay and blockchain offloading costs.

[0015] In addition, the effects according to the embodiments of the present disclosure mentioned above are not limited to the described contents, and may further include all effects predictable from the specification and drawings.

[0016] FIG. 1 is a block diagram of a dynamic redundancy-aware blockchain-based partial computation offloading system for a metaverse in an in-network computing environment according to one embodiment of the present disclosure.

[0017] FIG. 2 is a diagram illustrating a distributed algorithm proposed to solve a partial computation offloading problem modeled in the form of an ordinal potential game accessible to multiple users on the user side according to one embodiment of the present disclosure.

[0018] FIG. 3 is a diagram showing an algorithm for training a deep double Q-network according to an embodiment of the present disclosure.

[0019] FIG. 4 is a flowchart illustrating an operation mechanism for each time slot in a system according to an embodiment of the present disclosure.

[0020] FIG. 5 is a diagram illustrating the interaction between an algorithm and a system model according to one embodiment of the present disclosure.

[0021] FIG. 6 is a diagram showing trends in system cost and system compensation for repetition steps according to various episodes according to an embodiment of the present disclosure.

[0022] FIG. 7 is a diagram comparing a reward function based on average cost according to an embodiment of the present disclosure with OPG and OPG-Ran agents.

[0023] FIG. 8 is a diagram illustrating the influence of computing task types on computational overhead according to an embodiment of the present disclosure, comparing average cost, average delay, and average reward.

[0024] FIG. 9 is a diagram illustrating the influence of the maximum redundancy factor for each time slot according to one embodiment of the present disclosure, comparing the average cost and average reward.

[0025] FIG. 10 is a diagram illustrating an evaluation of the average cost and average reward of the performance of each time slot for a set of various numbers of users according to an embodiment of the present disclosure.

[0026] Throughout this disclosure, the same reference numerals denote the same components. This disclosure does not describe all elements of the embodiments, and any content that is common in the technical field to which this disclosure pertains or that overlaps between embodiments is omitted. The terms "part, module, element, block" used in the specification may be implemented in software or hardware, and depending on the embodiments, multiple "parts, modules, elements, blocks" may be implemented as a single component, or a single "part, module, element, block" may include multiple components.

[0027] Throughout the specification, when a part is said to be "connected" to another part, this includes not only direct connection but also indirect connection, and indirect connection includes connection via a wireless communication network.

[0028] Additionally, when a part is said to "include" a component, this does not mean that it excludes other components, but rather that it may include other components, unless otherwise specifically stated.

[0029] Throughout the specification, when we say that an element is "on" another element, this includes not only cases where the element is in contact with the other element, but also cases where another element exists between the two elements.

[0030] The terms first, second, etc. are used to distinguish one component from another, and the components are not limited by the aforementioned terms.

[0031] Singular expressions include plural expressions unless the context clearly indicates otherwise.

[0032] The identification codes for each step are used for convenience of explanation and do not describe the order of each step. Each step may be performed in a different order than specified unless the context clearly indicates a specific order.

[0033] The operating principle and embodiments of the present disclosure are described below with reference to the attached drawings.

[0034] As used herein, the term "device according to the present disclosure" encompasses a variety of devices capable of performing computational processing and providing results to a user. For example, the device according to the present disclosure may include a computer, a server device, and a portable terminal, or may be any one of them.

[0035] Here, the computer may include, for example, a notebook, desktop, laptop, tablet PC, slate PC, etc. equipped with a web browser.

[0036] The above server device is a server that processes information by communicating with an external device, and may include an application server, a computing server, a database server, a file server, a game server, a mail server, a proxy server, a web server, etc.

[0037] The above portable terminal may include, for example, a wireless communication device that ensures portability and mobility, and may include all kinds of handheld-based wireless communication devices such as a PCS (Personal Communication System), GSM (Global System for Mobile communications), PDC (Personal Digital Cellular), PHS (Personal Handyphone System), PDA (Personal Digital Assistant), IMT (International Mobile Telecommunication)-2000, CDMA (Code Division Multiple Access)-2000, W-CDMA (W-Code Division Multiple Access), WiBro (Wireless Broadband Internet) terminal, smart phone, etc., and wearable devices such as a watch, ring, bracelet, anklet, necklace, glasses, contact lens, or head-mounted device (HMD).

[0038] The artificial intelligence-related functions according to the present disclosure are operated via a processor and memory. The processor may be comprised of one or more processors. In this case, one or more processors may be a general-purpose processor such as a CPU, an AP, a DSP (Digital Signal Processor), a graphics-only processor such as a GPU or a VPU (Vision Processing Unit), or an artificial intelligence-only processor such as an NPU. One or more processors control the processing of input data according to predefined operating rules or artificial intelligence models stored in memory. Alternatively, if one or more processors are artificial intelligence-only processors, the artificial intelligence-only processor may be designed with a hardware structure specialized for processing a specific artificial intelligence model.

[0039] The predefined operation rules or artificial intelligence models are characterized by being created through learning. Here, being created through learning means that the basic artificial intelligence model is learned by a learning algorithm using a plurality of learning data, thereby creating a predefined operation rules or artificial intelligence model set to perform a desired characteristic (or purpose). This learning may be performed on the device itself on which the artificial intelligence according to the present disclosure is performed, or may be performed through a separate server and / or system. Examples of the learning algorithm include, but are not limited to, supervised learning, unsupervised learning, semi-supervised learning, or reinforcement learning.

[0040] An artificial intelligence model may be composed of multiple neural network layers. Each of the multiple neural network layers has multiple weight values, and performs neural network operations through operations between the operation results of the previous layer and the multiple weights. The multiple weights of the multiple neural network layers may be optimized based on the learning results of the artificial intelligence model. For example, the multiple weights may be updated so that the loss value or cost value obtained from the artificial intelligence model is reduced or minimized during the learning process. The artificial neural network may include a deep neural network (DNN), and examples thereof include, but are not limited to, a convolutional neural network (CNN), a deep neural network (DNN), a recurrent neural network (RNN), a restricted boltzmann machine (RBM), a deep belief network (DBN), a bidirectional recurrent deep neural network (BRDNN), or deep Q-networks.

[0041] According to an exemplary embodiment of the present disclosure, a processor can implement artificial intelligence. Artificial intelligence refers to a machine learning method based on an artificial neural network that imitates human neurons (biological neurons) to enable machines to learn. Artificial intelligence methodologies can be categorized into supervised learning, in which input data and output data are provided together as training data depending on the learning method, so that the solution (output data) to the problem (input data) is determined; unsupervised learning, in which only input data is provided without output data, so that the solution (output data) to the problem (input data) is not determined; and reinforcement learning, in which a reward (Reward) is provided from an external environment whenever an action (Action) is taken in the current state (State), and learning is performed in a direction to maximize this reward. In addition, artificial intelligence methodologies can be categorized according to the architecture of the learning model. The architectures of widely used deep learning technologies can be categorized into convolutional neural networks (CNNs), recurrent neural networks (RNNs), transformers, and generative adversarial networks (GANs).

[0042] The present device and system may include an artificial intelligence model. The artificial intelligence model may be a single artificial intelligence model or may be implemented as multiple artificial intelligence models. The artificial intelligence model may be composed of a neural network (or artificial neural network) and may include statistical learning algorithms that mimic biological neurons in machine learning and cognitive science. A neural network may refer to a model in general that has problem-solving capabilities by changing the binding strength of synapses through learning, formed by artificial neurons (nodes) that form a network by combining synapses. The neurons of the neural network may include a combination of weights or biases. The neural network may include one or more layers composed of one or more neurons or nodes. For example, the device may include an input layer, a hidden layer, and an output layer. The neural network constituting the device can infer a desired result (output) from an arbitrary input (input) by changing the weights of neurons through learning.

[0043] The processor can create a neural network, train (or learn) a neural network, perform a calculation based on received input data, generate an information signal based on the calculation result, or retrain the neural network. The models of the neural network can include various types of models such as CNN (Convolution Neural Network) such as GoogleNet, AlexNet, VGG Network, R-CNN (Region with Convolution Neural Network), RPN (Region Proposal Network), RNN (Recurrent Neural Network), S-DNN (Stacking-based deep Neural Network), S-SDNN (State-Space Dynamic Neural Network), Deconvolution Network, DBN (Deep Belief Network), RBM (Restrcted Boltzman Machine), Fully Convolutional Network, LSTM (Long Short-Term Memory) Network, Classification Network, etc., but are not limited thereto. The processor can include one or more processors for performing calculations according to the models of the neural network. For example, the neural network can be a deep neural network. It may include a deep neural network.

[0044] Neural networks include CNN (Convolutional Neural Network), RNN (Recurrent Neural Network), perceptron, multilayer perceptron, FF (Feed Forward), RBF (Radial Basis Network), DFF (Deep Feed Forward), LSTM (Long Short Term Memory), GRU (Gated Recurrent Unit), AE (Auto Encoder), VAE (Variational Auto) Encoder), DAE (Denoising Auto Encoder), SAE (Sparse Auto Encoder), MC (Markov Chain), HN (Hopfield Network), BM (Boltzmann Machine), RBM (Restricted Boltzmann Machine), DBN (Depp Belief Network), DCN (Deep Convolutional Network), DN (Deconvolutional Network), DCIGN (Deep Convolutional Inverse Graphics Network), Generative Adversarial Network (GAN), Liquid State Machine (LSM), Extreme Learning Machine (ELM), It will be understood by those skilled in the art that any neural network may be included, including but not limited to ESN (Echo State Network), DRN (Deep Residual Network), DNC (Differentiable Neural Computer), NTM (Neural Turning Machine), CN (Capsule Network), KN (Kohonen Network), and AN (Attention Network).

[0045] According to an exemplary embodiment of the present disclosure, the processor may be configured to perform a process for generating a CNN (Convolution Neural Network) such as GoogleNet, AlexNet, VGG Network, Region with Convolution Neural Network (R-CNN), Region Proposal Network (RPN), Recurrent Neural Network (RNN), Stacking-based deep Neural Network (S-DNN), State-Space Dynamic Neural Network (S-SDNN), Deconvolution Network, Deep Belief Network (DBN), Restrcted Boltzman Machine (RBM), Fully Convolutional Network, Long Short-Term Memory (LSTM) Network, Classification Network, Generative Modeling, eXplainable AI, Continual AI, Representation Learning, AI for Material Design, BERT, SP-BERT, MRC / QA for natural language processing, Text Analysis, Dialog System, GPT-3, GPT-4, Visual Analytics for vision processing, Visual Understanding, Video Synthesis, ResNet for data intelligence, Anomaly Detection, Prediction, Time-Series Forecasting, Various artificial intelligence structures and algorithms can be used, including but not limited to Optimization, Recommendation, and Data Creation.

[0046] Hereinafter, embodiments of the present disclosure will be described in detail with reference to the attached drawings.

[0047] To improve task offloading and resource allocation capabilities in existing services, we focused on applying Mobile Edge Computing (MEC) and Fog Computing to various applications.

[0048] However, it has been difficult to meet the requirements of the application because the Metaverse-based service requires very computationally intensive tasks such as rendering augmented reality (AR) and virtual reality (VR) on user equipment (UE), activating artificial intelligence for non-user characters (NPCs), and managing data for large numbers of users.

[0049] To address this, the present disclosure discloses the application of Computing In Network (COIN) and Blockchain technology, which are shared environments in which computing resources and services are connected through a network to provide complex services to users in a Metaverse service.

[0050] By incorporating the Computing In Network (COIN) paradigm, data reliability and security are guaranteed during offloading and resource allocation tasks. This facilitates resource and service exchange between user terminals and server nodes, while also reducing the cost of offloading AI-based non-user characters (NPCs). This process requires balancing memory, resource costs, and trust within the metaverse service.

[0051] In particular, the present disclosure discloses a method and system for optimizing offloading for applications and services in an in-network computing (COIN) environment, namely, a dynamic redundancy-aware blockchain-based partial computation offloading method and system for the metaverse within an in-network computing (COIN) environment.

[0052] In other words, the present disclosure relates to a method for effectively performing partial computation offloading (PCO) based on a blockchain to utilize unused network resources when computationally intensive applications and services such as a metaverse are built and provided in an in-network computing (COIN) environment, and more specifically, to a method and algorithm for optimizing offloading by formulating a blockchain redundancy factor (BRF) and partial computation offloading (PCO) problem by minimizing computational costs and maximizing incentives within a metaverse-based system, and then converting the problem into a real-time partial computation offloading (PCO) problem based on temporal correlation and a Markov decision process-based blockchain redundancy factor (BRF) problem, and designing a distributed algorithm for Nash equilibrium for the real-time partial computation offloading (PCO) problem and a double-deep Q-network-based algorithm for the blockchain redundancy factor (BRF) problem.

[0053] In this specification, a system and method are disclosed for significantly reducing cost overhead, providing higher rewards, and achieving optimal convergence in a few training episodes by defining a joint blockchain redundancy factor (BRF) and partial computation offloading (PCO) problem to minimize computational cost, maximize incentives, and satisfy latency and blockchain offloading constraints using a method and system according to the present disclosure, transforming the problem into two subproblems: real-time partial computation offloading (PCO) and Markov Decision Process (MDP)-based blockchain redundancy factor (BRF), and then using a distributed algorithm for the partial computation offloading (PCO) problem and a Deep Double Q-network-based algorithm (DDQA) for the blockchain redundancy factor (BRF) problem.

[0054] Meanwhile, in providing an optimal blockchain redundancy factor (BRF), the present disclosure optimizes data replication by considering dynamic factors such as cross-share communication overhead, shard allocation, blockchain costing, and overall offloading cost, and provides a criterion for users to make optimal offloading decisions.

[0055] In this context, the present disclosure presents a dynamic redundancy-aware blockchain-based partial computation offloading (PCO) for metaverse service operations in an in-network computing framework, where a blockchain redundancy factor (BRF) is periodically updated based on demand calculations, cost, and price constraints to support PCO users.

[0056] Through this feature, users who derive optimal Blockchain Redundancy Factors (BRFs) can make informed decisions based on the information provided by this disclosure to perform tasks within the metaverse, either locally or remotely.

[0057] In addition, the dynamic redundancy-aware blockchain-based partial computation offloading method and system for a metaverse in an in-network computing environment according to an embodiment of the present disclosure can provide an environment in which, when configuring a server and a network to improve the QoS of a metaverse service, three layers are arranged, and high-performance computation and storage devices such as in-network computing nodes and cloudlets are arranged therein to perform conventional traditional task offloading, as well as perform optimal offloading through an in-network computing agent and provide an scalable network and ultra-low-latency service.

[0058] In addition, the dynamic redundancy-aware blockchain-based partial computation offloading method and system for a metaverse in an in-network computing environment according to an embodiment of the present disclosure can maximize incentives in a dynamic metaverse service composed of EINs and FINs and present the joint blockchain redundancy factor (BRF) and partial computation offloading (PCO) problems that may occur when performing in-network computing-based offloading, thereby minimizing overhead of computation execution costs incurred when a user performs a task, while maximizing incentives and satisfying constraints on delay and blockchain offloading price.

[0059] As mentioned above, the shared blockchain redundancy factor (BRF) represents redundancy within a blockchain. Within a blockchain, data is distributed and this distributed storage can enhance data stability and reliability through redundancy. The shared blockchain redundancy factor (BRF) is an indicator of how many nodes within a blockchain network data is redundantly stored. In other words, a high shared blockchain redundancy factor (BRF) can enhance data security, but it can also impact the capacity and performance of the blockchain network.

[0060] Meanwhile, the Partial Computation Offloading (PCO) problem, related to "Byzantine Fault Tolerance," a consensus algorithm in distributed computing environments, represents the challenge of reaching reliable consensus among nodes in a distributed system. It addresses issues arising from malicious or faulty nodes on the network, and addressing these issues is a key element in enhancing the security and reliability of distributed ledger technologies such as blockchain.

[0061] FIG. 1 is a block diagram of a dynamic redundancy-aware blockchain-based partial computation offloading system for a metaverse in an in-network computing environment according to one embodiment of the present disclosure.

[0062] Referring to FIG. 1, an in-network computing (COIN) system according to one embodiment of the present disclosure may include a UE (User Equipment) (10) and an in-network computing (COIN) device that operates a metaverse service.

[0063] Here, the in-network computing (COIN) device may include a controller (400) that performs the role of a plurality of layers and an in-network computing agent (COIN agent).

[0064] The multiple layers may include, for example, a first layer (100), a second layer (200), a third layer (300), etc.

[0065] For convenience of explanation, the first layer (100) is named as the FIN (Fog COIN nodes) layer, the second layer (200) is named as the EIN (Edge COIN nodes) layer, and the third layer (300) is named as the blockchain layer.

[0066] According to FIG. 1, in order to perform traditional task offloading, user requests collected through user terminals (10) can be received in batches by a base station (110) of a FIN layer (100) to perform tasks within the metaverse.

[0067] User data can then be moved to in-network computing nodes (COIN nodes) (120) within the FIN layer (100).

[0068] In-network computing nodes (COIN nodes) (120) can transmit user data to a cloudlet (130). The user data transmitted in this manner can undergo task offloading in the cloudlet (130).

[0069] User data can be moved to the EIN layer (200) via the FIN layer (100).

[0070] After receiving user data as in-network computing nodes (COIN nodes) (210) within the EIN layer (200), the user data is transmitted to a cloud data center (220), and task offloading can be performed in the cloud data center (220).

[0071] To perform task offloading according to an embodiment of the present disclosure, as described above, user requests collected through user terminals (10) are collectively received by a base station (110) at the FIN layer (100) to perform tasks within the metaverse, and then user data is moved to the EIN layer (200). Thereafter, task offloading may be performed based on an optimal offloading decision calculated by the controller (400).

[0072] At this time, the controller (400) can perform dynamic redundancy-aware blockchain-based partial computation offloading (PCO) by considering and calculating a blockchain redundancy factor (BRF) based on the algorithm of FIGS. 2 and 3 described below, for example, to ensure the privacy and stability of user data received through the blockchain layer (300).

[0073] And in this disclosure, due to the interaction between partial computation offloading (PCO) and blockchain redundancy factor (BRF) in various time slots and the absence of user request transition probability, the problem can be defined by dividing it into the partial computation offloading (PCO) problem and the blockchain redundancy factor (BRF) problem to deal with this.

[0074] FIG. 2 is a diagram illustrating a distributed algorithm proposed to solve a partial computation offloading (PCO) problem modeled in the form of an ordinal potential game (ODG) that can be accessed by multiple users on the user side according to an embodiment of the present disclosure.

[0075] Referring to the algorithm in Figure 2, to address the multi-user partial computation offloading (PCO) problem and ensure mutual user satisfaction, offloading decisions are first initialized based on computational efficiency, thereby inducing rapid convergence to an optimal solution. In the case of remote offloading, the aforementioned FIN and EIN efficiencies can be calculated as needed.

[0076] Afterwards, to calculate the partial computation offloading (PCO) of each user and derive the optimal partial computation offloading (PCO) decision, constraints 26a, 26b, 26f and 26h are solved in an iterative-execution loop, and then the user can calculate the inference and forwarding rates for each subtask and define the strategy space.

[0077] In the above, constraints 26a, 26f, and 26h can guide optimal partial computation offloading (PCO) decisions to minimize an objective function based on delay and power consumption. Thereafter, the user terminal (10) can request an update from the controller (400). If the user terminal (10) does not receive an update, the decision can be maintained. Finally, the user terminal (10) can offload the task if it receives an END message.

[0078] FIG. 3 is a diagram showing an algorithm for training a deep double Q-network (DQN) according to an embodiment of the present disclosure.

[0079] According to the algorithm in Fig. 3, to solve the blockchain redundancy factor (BRF) problem modeled using a Markov decision process, an optimal redundancy level policy can be learned, for example, using a deep double Q-network (DQN).

[0080] In this disclosure, the model is evaluated against several common criteria, which may include Overestimation Penalty for Generalization (OPG) with full redundancy, OPG with randomly assigned redundancy elements (OPG-Random), Mobile Edge Computing (MEC) with full redundancy, and random-based offloading (Random) methods.

[0081] FIG. 4 is a flowchart illustrating an operation mechanism for each time slot in a system according to an embodiment of the present disclosure.

[0082] FIG. 4 may illustrate, for example, an operating method performed in a system or a controller (400) of the system.

[0083] In operation S110, the controller (400) can obtain a task operation request from the user terminal (10).

[0084] In operation S120, the controller (400) can calculate a task offloading policy based on the current network status.

[0085] In operation S130, the controller (400) can determine whether compute locally is possible.

[0086] In operation S140, if local operation is not possible as a result of the judgment in operation S130, the controller (400) can determine an offloading destination based on the cost and queue status.

[0087] In operation S150, the controller (400) can determine the offloading destination through BC based on cost and queue status. Here, BC can represent, for example, FIN or / and EIN.

[0088] In operation S160, the controller (400) can determine BC full redundancy.

[0089] In operation S170, the controller (400) can calculate optimal redundancy if the result of operation S160 is not BC full redundancy.

[0090] In operation S180, the controller (400) can offload task execution to BC, i.e., FIN or EIN, if the result of the judgment in operation S160 is BC full redundancy or / and the optimal redundancy is calculated in operation S170.

[0091] In operation S190, if the controller (400) determines that local operation is possible based on the S130 operation judgment result, local execution can be performed.

[0092] Referring to FIG. 4, when a task is requested from a user terminal (10), the controller (400) calculates a task offloading policy based on the current network status, and if the calculation is performed locally, it can be executed on the corresponding layer. Conversely, if the calculation is not performed locally, the controller (400) can determine the offloading destination among the FIN (100), EIN (200), and blockchain (300) layers based on cost and queue status.

[0093] At this time, depending on whether the blockchain is aware of redundancy, a step of calculating the optimal redundancy factor is performed, and work offloading can be performed to the FIN layer (100) or EIN layer (200).

[0094] FIG. 5 is a diagram illustrating the interaction between an algorithm and a system model according to one embodiment of the present disclosure.

[0095] Referring to FIG. 5, in operation S210, the controller (400) may coordinate the user terminal (10) to perform a partial computation offloading (PCO) game to resolve the partial computation offloading (PCO) policy. In the above, the partial computation offloading game may be based on, for example, Algorithm 1 of FIG. 2. Meanwhile, operation S210 may be performed, for example, at the start of time slot t.

[0096] In operation S220, the controller (400) may execute user tasks (jobs) based on a partial computation offloading (PCO) policy. Operation S220 may be performed, for example, during time slot t.

[0097] In operation S230, the controller (400) can determine whether training has been performed.

[0098] In operation S240, if the controller (400) is trained, it can tune the training algorithm for BC update and deep double Q-net (DDQN) training. At this time, the training algorithm illustrated in FIG. 3 can be used.

[0099] In operation S250, the controller (400) can execute an inference algorithm for BC update if it is not trained.

[0100] The above operations S230 to S250 may be performed, for example, at the end of time slot t.

[0101] Referring to FIG. 5, when time slot t starts, the controller (400) orchestrates user terminals to perform partial computation offloading (PCO) as in the algorithm illustrated in FIG. 2, and then, based on the partial computation offloading (PCO) policy, after the user's metaverse task requests are performed, the task of the controller (400) may vary depending on whether training is performed.

[0102] If trained, the controller (400) can tune the training algorithm illustrated in FIG. 3 for training the Deep Double Q-Net (DDQN) and updating the blockchain.

[0103] On the other hand, if not trained, the controller (400) can perform the inference algorithm of the Deep Double Q-Net (DDQN) to update the blockchain.

[0104] FIG. 6 is a diagram showing trends in system cost and system compensation for repetition steps according to various episodes according to an embodiment of the present disclosure.

[0105] Referring to Figure 6, the average cost and reward in various training episodes are evaluated compared to a common baseline, and the results show that the proposed model consistently has the lowest cost, followed by the Random, Mobile Edge Computing (MEC), OPG, and OPG-Ran models.

[0106] The average cost on the left side of Fig. 6 shows that the proposed method has been reduced by more than 100% compared to the random method, and the average reward on the right side of Fig. 6 also shows that the proposed method has achieved a system reward that is 99% higher than that of other methods.

[0107] FIG. 7 is a diagram comparing a reward function based on average cost according to an embodiment of the present disclosure with OPG and OPG-Rand agents.

[0108] Referring to Figure 7, it shows the ability of the agent to learn the optimal redundancy factor update for cost reduction and reward increase.

[0109] To this end, we analyzed the reward function related to the average cost, and chose the weighted sum of the cost difference and incentive as the quantitative value of the reward to minimize the cost and maximize the incentive as a performance indicator. At this time, the incentive is affected by the redundancy factor, and partial redundancy can reduce competition and decrease the overall reward.

[0110] Referring to Figure 7, the OPG method can maximize incentives by reducing costs from overall redundancy from a user perspective, but it also incurs higher costs. Therefore, the cost gap widens while the average cost increases.

[0111] The OPG-Ran method shows that random allocation of redundancy elements effectively reduces the overall system cost, as it optimally allocates redundancy elements to reduce delay and overall system cost.

[0112] In relation to the method proposed in this disclosure, first, the types of sub-tasks within the metaverse service were investigated, and then the tasks were classified as data-centric or computation-centric. As a result, it was found that out of the six tasks, three were classified as data-centric tasks and three were computation-centric tasks, and each was composed of four sub-tasks.

[0113] FIG. 8 is a diagram illustrating the influence of computing task types on computational overhead according to an embodiment of the present disclosure, comparing average cost, average delay, and average reward.

[0114] According to FIG. 8, in terms of cost, the method according to the present disclosure shows a cost reduction of 108% to 124% compared to the mobile edge computing (MEC) model for data-intensive tasks, and a cost reduction of 111% to 119% for compute-intensive tasks.

[0115] In addition, since achieving very low latency in the metaverse is an important indicator for ensuring QoS, the performance of the model was evaluated based on latency in this disclosure, and as a result, it was found that the method according to this disclosure shows a reduced latency of 109% to 124% for data-intensive tasks and 111% to 119% for compute-intensive tasks compared to mobile edge computing (MEC).

[0116] Additionally, considering that data-centric tasks primarily incur delays in communication transmission and computation-centric tasks require additional delays, it can be seen that, regardless of the task type, the controller (400) obtains about 100% more average compensation for the above criteria.

[0117] And according to the method according to the present disclosure, the impact of in-network computing (COIN) parameters on computational cost can be evaluated through analysis in four aspects including secondly, redundancy factor, number of users, and number of sub-tasks.

[0118] FIG. 9 is a diagram illustrating the influence of the maximum redundancy factor for each time slot according to one embodiment of the present disclosure, comparing the average cost and average reward.

[0119] Figure 9 shows the average cost and compensation per time slot when the method according to the present disclosure and the redundancy factor change.

[0120] More specifically, as the redundancy factor increases from, say, 5 to 30, the overall cost increases by about 82%, while the cost can be reduced by an average of about 59% compared to Mobile Edge Computing (MEC).

[0121] Furthermore, the OPG and OPG-Ran methods not only showed a significant increase in cost compared to the proposed method, but also showed a higher decrease in reward than the proposed method as the redundancy factor increased from, for example, 5 to 30, thereby highlighting the need for optimal estimation of the redundancy factor to improve cost-effectiveness and reward in blockchain-based offloading projects compared to the state-of-the-art OPG solution.

[0122] FIG. 10 is a diagram illustrating an evaluation of the average cost and average reward of the performance of each time slot for a set of various numbers of users according to an embodiment of the present disclosure.

[0123] Figure 10 shows the performance evaluation of the proposed method when the number of users changes.

[0124] It can be seen that the method according to the present disclosure consistently demonstrates cost efficiency compared to the OPG method as the number of users increases.

[0125] More specifically, the method according to the present disclosure reduces the cost by about 47% compared to the OPG method, and the OPG-Ran and random methods also follow the cost reduction, and it can be seen that the cost of mobile edge computing (MEC) remains stable as the number of users increases.

[0126] Additionally, in terms of reward, the method according to the present disclosure shows an improvement of about 64% for, for example, more than 20 users, which shows that the method according to the present disclosure is effective in reducing costs and maximizing rewards while determining the optimal offloading destination.

[0127] The present disclosure may also be implemented in a combined form of two or more of the above-described embodiments.

[0128] Meanwhile, the disclosed embodiments may be implemented in the form of a recording medium storing computer-executable instructions. The instructions may be stored in the form of program code, and when executed by a processor, may generate program modules to perform the operations of the disclosed embodiments. The recording medium may be implemented as a computer-readable recording medium.

[0129] Computer-readable storage media include all types of storage media that store instructions that can be deciphered by a computer. Examples include read-only memory (ROM), random access memory (RAM), magnetic tape, magnetic disks, flash memory, and optical data storage devices.

[0130] The disclosed embodiments have been described with reference to the attached drawings as described above. Those skilled in the art will understand that the present disclosure can be implemented in forms other than the disclosed embodiments without altering the technical spirit or essential features of the present disclosure. The disclosed embodiments are illustrative and should not be construed as limiting.

Claims

1. In a dynamic redundancy-aware blockchain-based partial computation offloading system for the metaverse in an in-network computing environment, User equipment sending a request for a metaverse operation; and Including an in-network computing device that operates a metaverse service, The above in-network computing device, With multiple layers, A controller that computes partial operation offloading and sends a request to determine a layer to perform the metaverse operation requested by the user device, The above plurality of layers include a first layer that receives user data including a work request transmitted from the user equipment and performs first work offloading or transmits it to a metaverse service network unit, The above plurality of layers includes a second layer that receives user data containing the work request through the first layer and performs the first work offloading or partial operation offloading, The above multiple layers include a third layer that ensures the stability of data containing user requests and guarantees privacy so as to determine the offloading policy for partial operation offloading and whether or not to recognize redundancy. The above controller, When a task is requested from a user terminal, the task offloading policy is calculated based on the current network status, If the above operation is performed locally, it is executed in the corresponding layer. A blockchain-based partial computation offloading system, configured to determine an offloading destination as one of the first layer, the second layer, and the third layer based on cost and queue status when the above computation is not performed locally.

2. In paragraph 1, The above user equipment, Ability to access the metaverse service built on the above network computing environment, A blockchain-based partial computation offloading system that performs the function of transmitting requests for metaverse tasks and sub-tasks from the above metaverse service to the server and network level.

3. In paragraph 1, The above controller, A blockchain-based partial computation offloading system, which performs the functions of preemptively determining the redundancy factor of a blockchain to determine an offloading method, defining metaverse task policies, measuring the redundancy factor of a blockchain where the requested metaverse task is stored, and controlling remote tasks and commands within the system.

4. In paragraph 1, The above first layer, A blockchain-based partial computation offloading system, which is an abstract or generalizable network layer, including a base station receiving user data transmitted from the user equipment, an in-network computing node capable of data communication with the controller or the base station, and a cloudlet performing first offloading.

5. In paragraph 4, The above in-network computing node, A blockchain-based partial computation offloading system, which is a node for a physical network device or a virtualized system, including a function of receiving an offloading command from the controller and transferring partial computation offloading to the user data and a function of transmitting data based on a user request to the cloudlet.

6. In paragraph 4, The above cloudlet is, A blockchain-based partial computation offloading system, which is a physical computation / storage device or virtual machine-based system capable of complex computation, including the ability to perform metaverse service requests transmitted from each of the above user devices.

7. In paragraph 1, The second layer above, A blockchain-based partial computation offloading system, which is an abstract or generalizable network layer, including an in-network computing node capable of data communication with the above controller or the first layer, and a cloud data center capable of performing the first offloading.

8. In paragraph 7, The above in-network computing node, A blockchain-based partial computation offloading system, which is a node for a physical network device or a virtualized system, which receives an offloading command from the controller and performs a function of transmitting partial computation offloading according to user data to the user equipment and a function of transmitting data based on a user request to the cloud data center.

9. In paragraph 7, The above cloud data center, A blockchain-based partial computation offloading system, which is a complex computationally capable physical computation / storage device or high-end virtual machine-based system that performs requests for metaverse operations transmitted from multiple user devices.

10. In paragraph 1, The third layer above is, A blockchain-based partial computation offloading system, which is a network layer having a plurality of distributed databases as components, which perform the function of encrypting user data containing metaverse operation requests of the user equipment and the function of storing transactions for requests of each user equipment.

11. A dynamic redundancy-aware blockchain-based partial computation offloading method for a metaverse within an in-network computing environment performed by a system, A step of receiving a request for a metaverse operation from a user device; A step of computing partial operation offloading through multiple layers; and A step of determining a layer to perform a metaverse task requested by the user device by the controller, The above plurality of layers include a first layer that receives user data including a work request transmitted from the user equipment and performs first work offloading or transmits it to a metaverse service network unit, The above plurality of layers includes a second layer that receives user data containing the work request through the first layer and performs the first work offloading or partial operation offloading, The above multiple layers include a third layer that ensures the stability of data containing user requests and guarantees privacy so as to determine the offloading policy for partial operation offloading and whether or not to recognize redundancy. The above controller, When a task is requested from a user terminal, the task offloading policy is calculated based on the current network status, If the above operation is performed locally, it is executed in the corresponding layer. A blockchain-based partial computation offloading method, configured to determine an offloading destination as one of the first layer, the second layer, and the third layer based on the cost and queue status when the above computation is not performed locally.

Images