Multi-resource Allocation Research Articles

Multi-agent deep reinforcement learning for dynamic reconfigurable shop scheduling considering batch processing and worker cooperation

Reconfigurable manufacturing system is considered as a promising next-generation manufacturing paradigm. However, limited equipment and complex product processes add additional coupled scheduling problems, including resource allocation, batch processing and worker cooperation. Meanwhile, dynamic events bring uncertainty. Traditional scheduling methods are difficult to obtain good solutions quickly. To this end, this paper proposes a multi-agent deep reinforcement learning (DRL) based method for dynamic reconfigurable shop scheduling problem considering batch processing and worker cooperation to minimize the total tardiness cost. Specifically, a dual-agent DRL-based scheduling framework is first designed. Then, a multi-agent DRL-based training algorithm is developed, where two high-quality end-to-end action spaces are designed using rule adjustment, and an estimated tardiness cost driven reward function is proposed for order-level scheduling problem. Moreover, a multi-resource allocation heuristics is designed for the reasonable assignment of equipment and workers, and a batch processing rule is designed to determine the action of manufacturing cell based on workshop state. Finally, a strategy is proposed for handling new order arrivals, equipment breakdown and job reworks. Experimental results on 140 instances show that the proposed method is superior to scheduling rules, genetic programming, and two popular DRL-based methods, and can effectively deal with various disturbance events. Furthermore, a real-world assembly and debugging workshop case is studied to show that the proposed method is applicable to solve the complex reconfigurable shop scheduling problems.

Online Multi-Resource Allocation for Network Slicing in 5G with Distributed Algorithms

With the increasing scale of mobile cellular network applications, 5G mobile network infrastructure provides customizable services to users in the form of network slices. How to effectively allocate existing resources in real dynamic networks with time-varying network utility is a key issue that previous work did not consider. This paper first initializes the multi-resource allocation problem of network slicing in an online manner, where the utility function is set to change over time. Therefore, we propose Metis, an online network slicing resource allocation framework that combines the time-varying nature of the network utility function given bandwidth and processing power constraints with the requirement of virtual network function isolation. The goal is to maximize the cumulative network utility in the long term and specify multiple resource allocation problems by utilizing concave optimization methods. In addition, a distributed algorithm based on the online alternating direction method of multipliers with regret optimization has been developed to achieve optimal resource allocation. Our mathematical analysis proves that Metis can provably converge to the optimal solution and the result of experiments demonstrates a steady state behavior of Metis, which converges in dynamic network settings.

Multi-resource allocation and care sequence assignment in patient management: a stochastic programming approach

To mitigate outpatient care delivery inefficiencies induced by resource shortages and demand heterogeneity, this paper focuses on the problem of allocating and sequencing multiple medical resources so that patients scheduled for clinical care can experience efficient and coordinated care with minimum total waiting time. We leverage highly granular location data on people and medical resources collected via Real-Time Location System technologies to identify dominant patient care pathways. A novel two-stage Stochastic Mixed Integer Linear Programming model is proposed to determine the optimal patient sequence based on the available resources according to the care pathways that minimize patients’ expected total waiting time. The model incorporates the uncertainty in care activity duration via sample average approximation.We employ a Monte Carlo Optimization procedure to determine the appropriate sample size to obtain solutions that provide a good trade-off between approximation accuracy and computational time. Compared to the conventional deterministic model, our proposed model would significantly reduce waiting time for patients in the clinic by 60%, on average, with acceptable computational resource requirements and time complexity. In summary, this paper proposes a computationally efficient formulation for the multi-resource allocation and care sequence assignment optimization problem under uncertainty. It uses continuous assignment decision variables without timestamp and position indices, enabling the data-driven solution of problems with real-time allocation adjustment in a dynamic outpatient environment with complex clinical coordination constraints.

A resource competition-based truthful mechanism for IoV edge computing resource allocation with a lowest revenue limit

Resource allocation in Internet of Vehicles (IoV) edge computing is currently a research hotspot. Existing studies focus on social welfare or revenue maximization. However, there is little research on lowest revenue guarantees, which is a problem of great concern to resource providers. This paper presents the innovative concept of the lowest revenue limit, which enables service providers to preset the revenue B and calculate whether the preset revenue can be achieved under the current supply and demand of resources through mechanism design. This approach is very friendly to service providers and can prevent low revenue and waste of resources. Specifically, we improved the ascending price auction mechanism so that it can be used for multi-resource allocation, the unit prices of different resources are calculated according to the intensity of competition among users, and the winning users and the payment are determined by eliminating users with low cost performance. Our mechanism is not sensitive to resource capacity, works well under deployment constraints in edge computing, and satisfies economic characteristics such as individual rationality and truthfulness. Compared with existing algorithms, our approach is shown to enable the service provider to obtain a higher revenue under a lower resource utilization.

MULTI-RESOURCE ALLOCATION AND LEVELING IN MULTI-PROJECT SCHEDULING PROBLEM WITH HYBRID-CHROMOSOME NON-DOMINATED SORTING GENETIC ALGORITHM II

Multi-resource allocation and leveling in multi-project (MR-AL-MP) scheduling refers to the attempt of producing a project schedule with minimum project duration and maximum resource utilization while complying with all precedence and resource availability constraints in a multi-project environment involving multiple resources. This study proposes a model that integrates both resource allocation and leveling models into a unified framework. This study develops a modified version of the Non-Dominated Sorting Genetic Algorithm II (NSGA-II), called Hybrid-Chromosome NSGA-II, as the optimization algorithm. For validation purposes, the performance of Hybrid-Chromosome NSGA-II is compared with two benchmark metaheuristic algorithms which are Multi-Objective Particle Swarm Optimization (MOPSO) and Multi-Objective Symbiotic Organisms Search (MOSOS) in optimizing a case study. It is shown that the proposed model and algorithm are able to produce a set of non-dominated solutions that represent the feasible trade-off relationships between the objectives. Furthermore, the Hybrid-Chromosome NSGA-II is superior to MOPSO and MOSOS in terms of the quality, spread, and diversity of the solutions.

FastAiAlloc: A real-time multi-resources allocation framework proposal based on predictive model and multiple optimization strategies

In cloud platforms, a common task is to allocate computational resources, e.g., memory and CPU, requested by applications and users. The allocation of these resources, which is an optimal load-balancing task, is considered an NP-Hard problem, being a challenging research area. There are many works proposed in the literature to address this problem. They use several strategies to deal with this optimization problem such as evolutionary algorithms, exact programming and also heuristics. However, some steps of the allocation process are not considered by these works, which are sometimes treated separately and not in an integrated manner. These steps include the applications and users resource consumption request profile as part of the optimization process and defining adequate metrics to check the optimized allocation. To integrate these steps, this work proposes a framework based on the following strategies, widely used in the literature: Genetic Algorithms (GA), Particle Swarm Optimization (PSO) and Linear Programming, besides our Heuristic approach. Furthermore, we restricted the resource allocation optimization to a real-time scenario, facilitating its use in an industrial process. In addition, Key Performance Indicators (KPIs) are proposed to carry out a comparative study in a public dataset. Our experiments showed that our proposed framework achieved better results than the baseline one, e.g., using a random allocation. In some cases, we observed an increase of almost 60% in performance compared with the baseline. In addition, when trying to balance memory and CPU consumption in the cluster, the linear approach performed the best, while the GA achieved the best result in allocating different user profiles across the cluster.

Fair Multi-Resource Allocation in Heterogeneous Servers With an External Resource Type

This paper considers the problem of fair allocation of multiple types of resources in heterogeneous servers, along with a resource type external to those servers. Our work is motivated by the need for fair multi-resource allocation in mobile edge computing (MEC), where the users must upload their tasks over a single dedicated wireless communication link that exists outside the computing servers. We propose a fair multi-resource allocation mechanism for this environment, termed Task Share Fairness with External Resource (TSF-ER), which finds the Kalai-Smorodinsky bargaining solution satisfying important fairness properties. We show that TSF-ER is envy-free, Pareto optimal, and strategy-proof, and it satisfies the property of sharing incentive. Large-scale simulation driven by Google and Alibaba cluster trace further shows that TSF-ER significantly outperforms the existing utilitarian, Nash social welfare maximizer, and egalitarian solutions, leading to fairer resource allocation while maintaining a high level of resource utilization.

Deep-Deterministic Policy Gradient Based Multi-Resource Allocation in Edge-Cloud System: A Distributed Approach

Edge Cloud (EC) empowers the beyond 5G (B5G) wireless networks to cope with large-scale and real-time traffics of Internet-of-Things (IoT) by minimizing the latency and providing compute power at the edge of the network. Due to a limited amount of resources at the EC compared to the back-end cloud (BC), intelligent resource management techniques become imperative. This paper studies the problem of multi-resource allocation (MRA) in terms of compute and wireless resources in an integrated EC and BC environment. Machine learning-based approaches are emerging to solve such optimization problems. However, it is challenging to adopt traditional discrete action space-based methods due to their high dimensionality issue. To this end, we propose a deep-deterministic policy gradient (DDPG) based temporal feature learning attentional network (TFLAN) model to address the MRA problem. TFLAN combines convolution, gated recurrent unit and attention layers together to mine local and long term temporal information from the task sequences for excellent function approximation. A novel heuristic-based priority experience replay (hPER) method is formulated to accelerate the convergence speed. Further, a pruning principle helps the TFLAN agent to significantly reduce the computational complexity and balance the load among base stations and servers to minimize the rejection-rate. Lastly, data parallelism technique is adopted for distributed training to meet the needs of a high-volume of IoT traffic in the EC environment. Experimental results demonstrate that the distributed training approach suites well to the problem scale and can magnify the speed of the learning process. We validate the proposed framework by comparing with five state-of-the-art RL agents. Our proposed agent converges fast and achieves up to 28% and 72% reduction in operational cost and rejection-rate, and achieves up to 32% gain in the quality of experience on average, compared to the most advanced DDPG agent.

Knowledge-based flexible resource allocation optimisation strategy for multi-tenant radio access network slicing in 5G and B5G

Formalisation of the network slice as a resource allocation unit is considered a promising aspect that enables scalable and flexible resource allocation among many tenants in 5G and beyond 5G (B5G) communication networks. However, the user traffic has to be passed through the central administration for processing, which leads to latency problems. To solve this problem, recent research works have suggested fixed central-to-edge resource allocation ratios as per the service type. However, this approach leads to over-provisioning of some resources. This paper provides a flexible resource allocation approach for 5G slice networks operating in a heterogeneous environment with multiple tenants and tiers. A radial basis-neural network is used to convert abstract specifications of simulation activities into precise resource needs, and then a genetic algorithm-based flexible multi-resource allocation scheme is proposed, where a versatile optimisation framework is used. The results show that the proposed approach outperforms such existing schemes.

Efficient and Fair Multi-Resource Allocation in Dynamic Fog Radio Access Network Slicing

Future wireless networks should meet heterogeneous service requirements of diverse applications, including interactive multimedia, augmented reality, and autonomous driving. The fog radio access network (Fog-RAN) is a novel architecture that enables efficient and flexible allocation of network resources to end users. However, guaranteeing application-specific service requirements while maximizing resource utilization is an open challenge in Fog-RANs. This article proposes a multiresource Fog-RAN slicing scheme that maximizes network resource utilization and satisfies important economic properties: Pareto-optimality, envy freeness, and sharing incentive. The proposed solution considers both heterogeneous resources (i.e., bandwidth, storage, and computing) and the different service levels defined in 5G networks. Accordingly, a two-level resource scheduling mechanism is devised to jointly allocate Fog-RAN resources to slices in two stages: 1) a broker allocates resources to slices at fog nodes over a given time window and 2) a slice hypervisor then allocates slice-specific resources at each fog node to users with a much shorter time scale. An extensive evaluation based on real-world data sets demonstrates that the proposed solution significantly increases the monetary gain of service providers, namely, by 32%–60% compared to the state of the art, including dynamic hierarchical resource allocation and dynamic slicing with proportional allocation.

Analyzing QoS factor in 5 G communication using optimized data communication techniques for E-commerce applications

The fifth generation (5 G) communication is characterized using gigabyte data speeds and ultra-low latency features. Therefore, service providers and mobile networks exploit this technology for real-time applications, including E-commerce. The peak data request in an E-commerce application/ website increases with consumer density and demands. The prime requirement in such environments is data optimization for controlling network traffic and sustaining Quality of Service (QoS). For providing a stabilized data handling procedure for E-commerce applications, this article introduces a Contingent Data Optimization Technique (CDOT). This technique provides multi-resource allocation rather than data compression, depending on the peak utilization. The application-specific data generation and communication are identified for network radio-resource allocation that satisfies the 5 G compliance. The decisions on application data peak and distributions are assured using federated learning. In this process, the network devices are monitored for the least and peak data movements for which the radio resources that deform stagnancy are allocated. The federated learning tunes the allocation recommendation using the resource to data mapping during the peak data span. This technique's performance is validated using data handling rate, application response, delay, backlogs, and peak utilization.

Extended efficiency and soft-fairness multiresource allocation in a cloud computing system

Extended efficiency and soft-fairness multiresource allocation in a cloud computing system

Fairness constraint efficiency optimization for multiresource allocation in a cluster system serving internet of things

SummaryMassive, diverse, and high‐frequency Internet of Things (IoT) applications pose challenges to the operation of cluster systems that serve it. Fair and efficient multidimensional resource allocation is of great significance to the sustainable operation of these systems. However, most of the existing cluster multiresource allocation optimization researches focus too much on the fairness of resource allocation and ignore the efficiency. The unbalanced use of multidimensional system resources reduces the effective utilization of system resources, which seriously affects the service quality of IoT applications. In this paper, we define the multiresource fair and efficient sharing optimization as a fairness‐constrained efficiency optimization problem, which is from dynamics, discrete resources, and heterogeneous perspectives according to the characteristics of cluster system in practical. Moreover, we present a dynamic efficiency‐aware multiresource fair allocation algorithm, DEF, which can improve the ability of the cluster system to serve diverse IoT applications. In the algorithm, large jobs schedule to the servers that expect the least remaining resources. Simulations performed using Google cluster‐usage traces show that DEF can improve system resource utilization and guarantee the fairness of sharing among users.

Multi resource allocation with partial preferences

We provide efficient, fair, and non-manipulable mechanisms for the multi-type resource allocation problems (MTRAs) and multiple assignment problems where agents have partial preferences over bundles consisting of multiple divisible items. We uncover a natural reduction from multiple assignment problems to MTRAs, which preserves the properties of MTRA mechanisms. We extend the well-known random priority (RP) and probabilistic serial (PS) mechanisms to MTRAs with partial preferences as multi-type PS (MPS) and multi-type RP (MRP) and propose a new mechanism, multi-type general dictatorship (MGD), which combines the ideas of MPS and MRP. We show that for the unrestricted domain of partial order preferences, unfortunately, no mechanism satisfies both sd-efficiency and sd-envy-freeness, even as they each satisfy different weaker notions of the desirable properties of efficiency, fairness, and non-manipulability we consider. Notwithstanding this impossibility result, our main message is positive: When agents' preferences are represented by acyclic CP-nets, MRP satisfies ex-post-efficiency, sd-strategyproofness, and upper invariance, while MPS satisfies sd-efficiency, sd-envy-freeness, ordinal fairness, and upper invariance, recovering the properties of RP and PS; the MGD satisfies sd-efficiency, equal treatment of equals, and decomposability under the unrestricted domain of partial preferences. We introduce a natural domain of bundle net preferences, which generalizes previously studied domain restrictions of partial preferences for multiple assignment problems and is incomparable to the domain of acyclic CP-nets. We show that MRP and MPS satisfy all properties of the RP and PS under bundle net preferences as well.

Distributed Algorithms for Multi-Resource Allocation

Novel network infrastructures require the distribution of computing and network resource control to meet stringent requirements in terms of latency, reliability and bitrate. 5G systems bring a key novelty in systems design that it the ‘network slice’as a new resource provisioning entity. A network slice is meant to serve end-to-end services as a composition of different network and system resources as radio, link, storage and computing resources. Conventionally, each resource is managed by a distinct decision-maker, platform, provider, orchestrator or controller. Naturally, centralized slice orchestration approaches are proposed in the literature, where a multi-domain orchestrator allocates the resources, for instance using a multi-resource allocation rule. Nonetheless, while simplifying the algorithmic approach, centralization can come at the expense of scalability and performance. In this article, we propose new ways to distribute the slice multi-resource resource allocation problem, using cascade and parallel resource allocations that are functionally compatible with novel software platforms. We also show how to adapt the proposed algorithms to make them able to guarantee service level agreements on the minimum resource needed, and to take into account deadline priority policy scheduling. We provide an exhaustive analysis of the advantages and disadvantages of the different approaches, including a numerical analysis for a realistic setting.

A Truthful Double Auction Mechanism for Multi-Resource Allocation in Crowd Sensing Systems

As a novel sensing paradigm, crowd sensing systems have gained great attention and been widely adopted in the environmental monitoring and calculation areas. In crowd sensing systems, mobile users provide their multiple resources to the requesters to execute tasks. Existing studies focus on the divisible task or one-to-one mapping for single resource allocation. However, this assumption does not hold for crowd sensing systems. Owing to the task attribute, some tasks cannot be divided into multiple parts to run on different devices. In addition, a high performance mobile device can execute multiple tasks simultaneously. We address the problem of multi-resource allocation in crowd sensing systems for the auction-based model considering many-to-one mapping for indivisible tasks, where many-to-one mapping allows one mobile device to provide multiple resources to execute one or more tasks. In this article, we study, for the first time to the best of our knowledge, a truthful mechanism that stimulates mobile users and requesters to declare their true values. We design a truthful double auction mechanism together with a payment scheme tailored to fit it that would help researchers understand how a truthful double auction mechanism can be designed. In addition, we prove that our proposed mechanism maintains budget-balance, individual rationality, and computational tractability. Furthermore, we analyze the approximation ratio of our proposed approximation algorithm. Experimental results demonstrate that our proposed mechanism has high computation efficiency and good performance.

Design and Analysis of an Efficient Multiresource Allocation System for Cooperative Computing in Internet of Things

By migrating tasks from the end devices to the edge or cloud, cooperative computing in the Internet of Things can support time-sensitive, high-dimensional, and complex applications while utilizing existing resources, such as the network bandwidth, computing resources, and storage capacity. How to design the multiresource allocation system efficiently is a significant research problem. In this article, we design a multiresource allocation system for cooperative computing in the Internet of Things based on deep reinforcement learning by redefining latency calculation models for communication, computation, and caching with the consideration of practical interference factors, such as the Gaussian noise and data loss. The proposed system uses actor–critic as the base model for rapidly approximating the optimal policy by updating parameters of the actor and critic in respective gradient directions. The balance control parameter is introduced to fit the desired learning rate and actual learning rate. At the same time, we use the method of double experience pool to limit the exploration direction of the optimal policy, which reduces the time complexity and space complexity of the problem solution and improves the adaptability and reliability of the scheme. Experiments have demonstrated that multiresource allocation algorithm based on deep reinforcement learning (DRL-MRA) performs well in terms of the average service latency under resource-constrained conditions, and the improvement is significant with the increase of network size.

Deep Reinforcement Learning for energy-aware task offloading in join SDN-Blockchain 5G massive IoT edge network

The Internet-of-Things (IoT) edge allows cloud computing services for topology and location-sensitive distributed computing. As an immediate benefit, it improves network reliability and latency by enabling data access and processing rapidly and efficiently near IoT devices. However, it comes with several issues stemming from the complexity, the security, the energy consumption, and the instability due to the decentralization of service localization. Furthermore, the multi-resource allocation and task scheduling make this task the furthest from being straightforward. Blockchain has been envisioned to enforce trustworthiness in diverse IoT environments. However, high latency and high energy costs are incurred to process IoT transactions. This paper introduces a novel Blockchain-based Deep Reinforcement Learning (DRL) approach to enable energy-aware task scheduling and offloading in an Software Defined Networking (SDN)-enabled IoT network. The Asynchronous Actor–Critic Agent (A3C) DRL-based policy achieves efficient task scheduling and offloading. The latter is in symbiosis with Proof-of-Authority Blockchain consensus to validate IoT transactions and blocks. By doing so, we improve reliability and low latency and achieve energy efficiency for SDN-enabled IoT networks. The A3C policy combined with the Blockchain is proved theoretically. Carried out experiments put forth that our approach offers 50% better energy efficiency, which outperforms traditional consensus algorithms, i.e., Proof of Work and PBFT, in terms of throughput and network latency and offers better scheduling performance.

Heterogeneous Multi-resource Allocation with Subset Demand Requests

We consider the problem of allocating multiple heterogeneous resources geographically and over time to meet demands that require some subset of the available resource types simultaneously at a specified time, location, and duration. The objective is to maximize the total reward accrued from meeting (a subset of) demands. We model this problem as an integer program, show that it is NP-hard, and analyze the complexity of various special cases. We introduce approximation algorithms and an extension to our problem that considers travel costs. Finally, we test the performance of the integer programming model in an extensive computational study.

An online fair resource allocation solution for fog computing

Fog computing is a complementary computing paradigm to the existing cloud computing. A fundamental problem of fog computing is how to allocate the computing resources of fog nodes when scheduling tasks that arrive in an online manner. Other than task completion speed metrics, fairness of resource allocation between competing users is also an important metric to consider. One such metric is Dominant Resource Fairness (DRF), a fairness scheme that guarantees four key qualities: incentivised sharing, strategy-proof, Pareto-efficiency, and envy free. This paper examines the multi-resource, multi-server, and heterogeneous task resource allocation problem from a DRF perspective. Four different types of tasks are considered: ordered/unordered and splittable/unsplittable. Three low complexity heuristics are proposed to maximise fairness between users. Results show that the proposed heuristics are at least comparable to three baseline scheduling algorithms in terms of task completion speed while achieving higher fairness between users.