Heuristic Allocation Research Articles

Dynamic resource allocation and scheduling for appointment-based systems with walk-ins

Effectively managing constrained and perishable capacity amid fluctuating demands is a common challenge in service industries, often addressed through appointment-based systems that incorporate walk-ins. Balancing appointments and walk-ins is crucial for enhancing system value and maintaining stability. In this article, we study an integrated decision problem for appointment scheduling and resource allocation across multiple facilities, considering stochastic demands from three segments: priority customers, walk-in customers, and appointment customers. We formulate the problem as a sequential scheduling-allocation challenge using stochastic dynamic programming. Leveraging the anti-multimodularity of the value function, we fully characterize the optimal dynamic resource allocation policy and its monotone structural properties under any appointment schedule. Given the complexity and intractability of optimal scheduling with dynamic resource allocation, we introduce two suboptimal scheduling policies—vertical scheduling and horizontal scheduling—and a heuristic allocation policy. Additionally, we create an Upper Bound (UB) for the original problem as a surrogate, establishing an asymptotically optimal UB for the scaled problem through its objective value and a corresponding lower bound through its solution. This UB solution serves as an efficient and effective heuristic, demonstrating superior performance compared with the combined performance of suboptimal scheduling policies with optimal dynamic resource allocation, especially as problem size increases. It also holds significant practical implications for integrated pooling systems.

Performance Modeling of Distributed Data Processing in Microservice Applications

Microservice applications are increasingly adopted in distributed data processing systems, such as in mobile edge computing and data mesh architectures. However, existing performance models of such systems fall short in providing comprehensive insights into the intricate interplay between data placement and data processing. To address these issues, this paper proposes a novel class of performance models that enables joint analysis of data storage access workflows, caching, and queueing contention. Our proposed models introduce a notion of access path for data items to model hierarchical data locality constraints. We develop analytical solutions to efficiently approximate the performance metrics of these models under different data caching policies, finding in particular conditions under which the underlying Markov chain admits a product-form solution. Extensive trace-driven simulations based on real-world datasets indicate that service and data placement policies based on our proposed models can respectively improve by up to 35% and 37% the average response time in edge and data mesh case studies compared to baseline resource allocation heuristics.

Load-balanced Routing Heuristics for Bandwidth Allocation of AVB Flow in TSN

Time-Sensitive Networking (TSN) is a new technology developed from Ethernet that guarantees deterministic transmission of various types of flows, such as Time-triggered (TT) flows and Audio-video-bridging (AVB) flows, in the same network. Currently, Time-aware Shaping (TAS) and Credit-based Shaping (CBS) are widely used for scheduling hybrid flows, where the credit value in CBS is directly linked to the logical bandwidth value idleSlope , which affects the real-time performance of AVB flows. However, as the network scale increases, existing bandwidth allocation methods result in higher computation times and lower allocation success rates. In this paper, the aforementioned problem is addressed by two-step: first, we design a load-balanced routing heuristic (LBRH) to improve the allocation success rate; second, we accelerate the CBS bandwidth allocation by using unified bandwidth allocation scheme, which integrates LBRH and further improves the allocation success rate. Performance evaluation in multiple test cases shows that LBRH can improve the success rate of bandwidth allocation, and the unified bandwidth allocation scheme integrating LBRH significantly reduces the overall execution time while improving the success rate of bandwidth allocation, which is more suitable for large-scale TSN networks with complex routing.

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.

Time-predictable task-to-thread mapping in multi-core processors

The performance of time-predictable systems can be improved in multi-core processors using parallel programming models (e.g., OpenMP). However, schedulability analysis of parallel applications is a big challenge due to their sophisticated structure. The common drawbacks of current task-to-thread mapping approaches in OpenMP are that they (i) utilize a global queue in the mapping process, which may increase contention, (ii) do not apply heuristic techniques, which may reduce the predictability and performance of the system, and (iii) use basic analytical techniques, which may cause notable pessimism in the temporal conditions. Accordingly, this paper proposes a task-to-thread mapping method in multi-core processors based on the OpenMP framework. The mapping process is carried out through two phases: allocation and dispatching. Each thread has an allocation queue in order to minimize contention, and the allocation and dispatching processes are performed using several heuristic algorithms to enhance predictability. In the allocation phase, each task-part from the OpenMP DAG is allocated to one of the allocation queues, which includes both sibling and child task-parts. A suitable thread (i.e., allocation queue) is selected using one of the suggested heuristic allocation algorithms. In the dispatching phase, when a thread is idle, a task-part is selected from its allocation queue using one of the suggested heuristic dispatching algorithms and then dispatched to and executed by the thread. The performance of the proposed method is evaluated under different conditions (e.g., varying the number of tasks and the number of threads) in terms of application response time and overhead of the mapping process. The simulation results show that the proposed method surpasses the other methods, especially in the scenario that includes overhead of the mapping. In addition, a prototype implementation of the main heuristics is evaluated using two kernels from real-world applications, showing that the methods work better than LLVM's default scheduler in most of the configurations.

Two-stage approximation allocation approach for real-time parking reservations considering stochastic requests and reusable resources

The digital economy has greatly impacted the traditional parking industry. This study focuses on a real-time parking reservation system that allocates available parking slots (supplies) to serve the parking requests (demands) of drivers within a certain area. The challenge lies in dealing with the randomness of requests with varying parking durations and the reusability of parking resources, which have not been well researched. To address this problem, we propose a stochastic optimization model and a data-driven two-stage approximation approach. The first stage generates a pre-allocation scheme as a basis, while the second stage dynamically adjusts the scheme based on observed requests and available parking slots to improve the system’s performance throughout a finite operational horizon. The study has three main contributions. Firstly, this paper proposes a data-driven nonparametric approach based on sample average approximation (SAA) to solve the stochastic optimization model. Secondly, it develops a stratified ranked set sampling (SRSS) method to ensure the quality of the sample set and proves that the sample set sampled by the SRSS method is unbiased. Finally, an evaluation method is proposed to establish a confidence interval for the expected revenue obtained by the data-driven method. Experimental results based on real parking demand data show that our approach outperforms both a commonly used heuristic allocation policy in the parking industry and the standard SAA approach with two popular sampling methods and an evaluation method based on Student’s T-test in the literature.

Deep Reinforcement Learning for One-Warehouse Multi-Retailer inventory management

The One-Warehouse Multi-Retailer (OWMR) system is the prototypical distribution and inventory system. Many OWMR variants exist, e.g. demand in excess of supply may be completely back-ordered, partially back-ordered, or lost. Prior research has focused on the study of heuristic reordering policies such as echelon base-stock levels coupled with heuristic allocation policies. Constructing well-performing policies is time-consuming and must be redone for every problem variant. By contrast, Deep Reinforcement Learning (DRL) is a general purpose technique for sequential decision making that has yielded good results for various challenging inventory systems. However, applying DRL to OWMR problems is nontrivial, since allocation involves setting a quantity for each retailer: The number of possible allocations grows exponentially in the number of retailers. Since each action is typically associated with a neural network output node, this renders standard DRL techniques intractable. Our proposed DRL algorithm instead inferences a multi-discrete action distribution which has output nodes that grow linearly in the number of retailers. Moreover, when total retailer orders exceed the available warehouse inventory, we propose a random rationing policy that substantially improves the ability of standard DRL algorithms to train good policies because it promotes the learning of feasible retailer order quantities. The resulting algorithm outperforms general-purpose benchmark policies by ∼1−3% for the lost sales case and by ∼12−20% for the partial back-ordering case. For complete back-ordering, the algorithm cannot consistently outperform the benchmark.

A Period Training Method for Heterogeneous UUV Dynamic Task Allocation

In the dynamic task allocation of unmanned underwater vehicles (UUVs), the schemes of UUVs need to be quickly reallocated to respond to emergencies. The most common heuristic allocation method uses predesigned optimization rules to iteratively obtain a solution, which is time-consuming. To quickly assign tasks to heterogeneous UUVs, we propose a novel task allocation algorithm based on multi-agent reinforcement learning (MARL) and a period training method (PTM). The period training method (PTM) is used to optimize the parameters of MARL models in different training environments, improving the algorithm’s robustness. The simulation results show that the proposed methods can effectively allocate tasks to different UUVs within a few seconds and reallocate the schemes in real time to deal with emergencies.

Optimal Traffic Load Allocation for Aloha-Based IoT LEO Constellations

The deployment of satellite networks is key to providing global wireless connectivity for the Internet of Things (IoT). In this line, we consider a cluster of IoT devices served by a constellation of low-Earth-orbit (LEO) satellites, while slotted Aloha is used as a medium-access-control (MAC) technique in the uplink. To characterize the channel, we employ an ON–OFF fading channel model that estimates the quality of the links between the cluster of IoT devices and the LEO satellites within the constellation, by taking into account their relative positions. Since each relative position of the constellation with respect to the cluster of IoT devices leads to a different throughput for a given traffic load, we propose a novel traffic load distribution strategy based on successive convex approximation (SCA) to maximize the system throughput. The method adequately allocates the traffic load among the different constellation positions with respect to the IoT cluster. Finally, the results show that the proposed method outperforms other recently proposed strategies based on heuristics for traffic load allocation, while it also achieves a stable nonzero throughput even for large traffic loads.

Underpinnings of User-Channel Allocation in Non-orthogonal Multiple Access for 5G

Non-orthogonal multiple access (NOMA) is a part of 5th generation (5G) communication systems. This article presents the underpinnings and underlying structures of the problem of NOMA user-channel allocation. Unlike the heuristics for NOMA user-channel allocation, the presented results are guaranteed to converge to a solution. In addition, the solutions are stable. More specifically, the deployment of results to cellular vehicular communication systems is shown as a use case of 5G technology in smart transport. Generally, the results apply to any NOMA system. Unlike the orthogonal frequency division multiple access resource allocation problem, the core matching is not the solution to NOMA resource allocation. The conditions under which the fix-point NOMA resource allocation is guaranteed to be stable from the viewpoint of both the base station and the NOMA users are described. In addition, relationships of NOMA user-channel resource allocation to game models and subgame perfect Nash equilibria are elucidated.

Steppe Generosity: Kinship, social reputations, and perceived need drive generous giving in a non-anonymous allocation game among Mongolian pastoral nomads

This study explores generosity among Mongolian pastoral nomads using a recipient identity-conditioned heuristic (RICH) allocation game in which study participants could allocate experimental funds among themselves and non-anonymous individuals in their local community. Allocation games were conducted with a sample of forty-six male and female pastoral household heads in Tosontsengel, Mongolia. Results indicate there were positive relationships between players' social reputations for being hardworking, skilled pastoralists, and generous and the amount of money they received from others. Players also gave more money to kin than non-kin. The results also indicate that players' allocation decisions were driven primarily by kinship and players' assessment of others' relative need.

Dynamic Control of Non‐Collaborative Workers When Reassignment Is Costly

Cross‐training is an important tool to improve the performance of manufacturing and service systems through dynamic worker allocation. This study investigates how cross‐training can be leveraged when worker collaboration is not possible and frequent worker reassignment is undesirable. We consider a tandem queueing system with finite buffers between tasks. We show that if each worker is the fastest at a different task, then the optimal policy is always dedicated, regardless of the magnitude of the reassignment costs. Otherwise, if the system is Markovian with two homogeneous tasks and a faster and a slower worker, we completely characterize how the profit‐optimal policy depends on the (constant) reassignment cost. We also prove that for any given reassignment cost, dedicated worker allocation policies are strictly suboptimal for large enough buffer sizes. Instead, the faster worker should move to the downstream task only if there are enough jobs in the intermediate buffer and return to the upstream task when the buffer is sufficiently empty. Furthermore, the benefit of task switching increases as the buffer size becomes larger. We use our theoretical results to develop worker allocation heuristics both for more general systems with two tasks and for systems with more tasks. Numerical experimentation provides insights on how the profit depends on the buffer sizes and reassignment costs, shows that policies that are optimal when workers are collaborative result in excessive switching for non‐collaborative workers, and indicates that our pick‐the‐best heuristics perform well in all settings.

Interference-Aware and Spectral-Efficient Resource Allocation Using Complete Graphs for the D2D Communication

This paper proposes a heuristic resource block allocation using complete graphs (abbreviated to RACG) mechanism for device-to-device (D2D) communication. RACG not only can reduce the cochannel interference but also can improve the overall system capacity. For RACG, we define two pieces of user equipment in the D2D communication as one D2D pair and use the interference between each D2D pair to establish an interference graph model, which is then used to identify all complete graphs with different numbers of vertices. The innovation of RACG is right that it begins with resource block (RB) allocation for the D2D pairs in the complete graph with the largest number of vertices and considers three additional parameters to determine the allocation sequence for the D2D pairs. The three additional parameters are the frequency at which D2D pairs appear in complete graphs, the interference level of a D2D pair to other D2D pairs, and the number of RBs required. Furthermore, RACG prioritizes the reuse of allocated RB chunks during RB allocation. When a D2D pair and other D2D pairs with allocated RBs do not have mutual interference, the allocated RB chunk that has the closest number of RBs to the number of RBs required is selected and reused. We conduct performance analysis and compare the proposed RACG with a resource allocation of degree-based greedy (RADG) approach. The performance measures include the RB usage rate and occupied rate in a frame.

Path Planning for Multi-Vehicle-Assisted Multi-UAVs in Mobile Crowdsensing

Due to the capability of fast deployment and controllable mobility, unmanned aerial vehicles (UAVs) play an important role in mobile crowdsensing (MCS). However, constrained by limited battery capacity, UAVs cannot serve a wide area. In response to this problem, the ground vehicle is introduced and used to transport, release, and recycle UAVs. However, existing works only consider a special scenario: one ground vehicle with multiple UAVs. In this paper, we consider a more general scenario: multiple ground vehicles with multiple UAVs. We formalize the multi-vehicle-assisted multi-UAV path planning problem, which is a joint route planning and task assignment problem (RPTSP). To solve RPTSP, an efficient multi-vehicle-assisted multi-UAV path planning algorithm (MVP) is proposed. In MVP, we first allocate the detecting points to proper parking spots and then propose an efficient heuristic allocation algorithm EHA to plan the paths of ground vehicles. Besides, a genetic algorithm and reinforcement learning are utilized to plan the paths of UAVs. MVP maximizes the profits of an MCS carrier with a response time constraint and minimizes the number of employed vehicles. Finally, performance evaluation demonstrates that MVP outperforms the baseline algorithm.

Capacity allocation in a service system with preferred service completion times

AbstractRetailers use different mechanisms to enable sales and delivery. A relatively new offering by companies is curbside pickup where customers purchase goods online, schedule a pickup time, and come to a pickup facility to collect their orders. To model this service structure, we consider a service system where each arriving job has a preferred service completion time. Unlike most service systems that operate on a first‐come‐first‐serve basis, the service provider makes a strategic decision for when to serve each job considering their requested times and the associated costs. For most of our results, we assume that all jobs must be served before or on their requested time period, and the jobs are handled in overtime when capacity is insufficient. Costs are incurred both for overtime and early service. We model this problem as a Markov decision process. For small systems, we show that optimal capacity allocation policies are of threshold type and provide additional structural results for special cases. Building on these results, we devise two capacity allocation heuristics that use a threshold structure for general systems. The computational results show that our heuristics find near‐optimal solutions, and dependably outperform the benchmark heuristics even in larger systems. We conclude that there is a considerable benefit in using our heuristics as opposed to a very greedy or a very prudent benchmark heuristic, especially when the early service costs are not prohibitively high and the service capacity is scarce or there are high volumes of customer arrivals. Our results also demonstrate that as the length of the customer order horizon increases, performance of all heuristics deteriorate but the benefits of using our threshold heuristic remain considerable. Finally, we provide guidelines to select an appropriate solution method considering the trade‐off between solution quality and computation time.

Ethical heuristics for pandemic allocation of ventilators across hospitals.

In response to the COVID-19 pandemic philosophers and governments have proposed scarce resource allocation guidelines. Their purpose is to advise healthcare professionals on how to ethically allocate scarce medical resources. One challenging feature of the pandemic has been the large numbers of patients needing mechanical ventilatory support. Guidelines have paradigmatically focused on the question of what doctors should do if they have fewer ventilators than patients who need respiratory support: which patient should get the ventilator? There is, however, an important higher level allocation problem. Namely, how are we to ethically distribute newly obtained ventilators across hospitals: which hospital should get the ventilator(s)? In this paper, we identify a set of principles for allocating newly obtained ventilators across hospitals. We focus particularly on low and middle income countries, who frequently have limited pre-existing intensive care capacity, and have needed to source additional ventilators. We first provide some background. Second, we argue that the main population healthcare aim during the COVID-19 pandemic should be to save the most lives. Next, we assess a series of potential heuristics or principles that could be used to guide allocation: allocation to the most densely populated cities, random allocation, allocation based on the ratio of patients to ICU personnel, prioritisation in terms of intrahospital mortality, prioritisation of younger populations, and prioritisation in terms of population mortality. We conclude by providing a plausible ranking of the principles, while noting a number of epistemological challenges, in terms of how they best further the aim of increasing the probability of saving the most lives.

A Heuristic Loss Allocation Method Based on Tracing Power Flow

This paper presents a new approach to solving the problem of loss allocation by assigning the losses to generators and consumers, which are the two non-regulated agents of the market. This proposed method suggests assigning the losses of a network line to the nodes which are connected to it as a function of the direction of the active power flow through the line. The nodes with connected loads will be assigned the losses of the lines carrying incoming active power flows at the node, while the nodes with generators connected will be assigned the losses of the lines carrying outgoing power flows. This allocation will be weighted according to the number of allocations carried out in the rest of the network nodes and also taking into consideration the power injected into the node. This heuristic method is developed and illustrated with simple examples, both the procedure as the algorithm. The method has been tested on the IEEE 30 bus test system and the results have been compared with other methods so as to show its effectiveness.

Can Adaptive Seriational Risk Parity Tame Crypto Portfolios?

As cryptocoins are not tied to fundamental values or to investor protection regulation, their price dynamics is unhinged in both directions. In institutional asset management of conventional asset classes, target volatility concepts and dynamic allocation heuristics are popular to improve the robustness of portfolios. Can similar techniques also be used to construct delevered and diversified portfolios of crypto assets? A robust candidate approach for allocation is Hierarchical Risk Parity (HRP), as it incorporates a filtered correlation structure and is less sensitive to noise than quadratic optimization, as shown in several studies. Recent publications have extended the concept of HRP in several directions. We compare some of these extensions to determine which variant is most useful for constructing crypto baskets. We find that a particular type of adaptive HRP strategy outperforms other extensions on a risk-adjusted basis, leading us to a deeper investigation of the changing nature of correlation structures between cryptos - both quantitatively and visually. We find that structural breaks in crypto correlations are prevalent and that the best-fitting hierarchical cluster representations change over time, which is only captured by distance matrix-based adaptive HRP approaches.

Coordinated global and private job-flow scheduling in grid virtual organizations

In this work, an approach for a preference-based job-flow scheduling in Grid virtual organizations is proposed and studied. Virtual organizations provide uniform rules of the resources sharing and consumption and should respect individual and common preferences of its stakeholders: users, resource providers and administrators. In most cases, a virtual organization’s stakeholders pursue contradictory market-based goals, which may be expressed in terms of time and cost. Still, mutually beneficial job-flow scheduling scenarios are possible in such multi-agent computing environments when the different optimization criteria are coordinated. For this purpose we consider different resources allocation heuristics and scenarios based on a simple linear combination of global (group) and private (user) job scheduling criteria. We study backfilling-based job queue scheduling procedure with a secondary preference-based optimization performed only on the resources allocation step. This approach allows us to respect backfilling-based jobs execution priorities together with more specific individual jobs criteria, including total execution cost or time minimization.

Modeling and simulation of hierarchical task allocation system for energy-aware HPC clouds

The paper presents a hierarchical approach to the problem of energy-aware task allocation in a distributed HPC system consisting of several computing clusters. The solution involves the control of clusters and the use of network traffic engineering methods to minimize overall energy consumption. The advantages of the hierarchical approach are twofold: first entities controlling parts of computing infrastructure may decide how much of their responsibility is delegated to the coordinator, next the computational complexity of related mathematical programming problems may be reduced, and their solution may be done in parallel. The mathematical programming problem of task allocation is defined in the centralized form first. After discussing its computation complexity and related organizational difficulties, it is decomposed to form a hierarchy. Finally, heuristics for allocation are proposed and verified via numerical simulation.