Decentralized Resource Allocation Research Articles

Distributed discrete‐time optimization algorithms with applications to resource allocation in epidemics control

SummaryThis work presents and analyzes 2 novel distributed discrete‐time nonlinear algorithms to solve a class of decentralized resource allocation problems. The algorithms allow an interconnected group of agents to collectively minimize a global cost function under inequality and equality constraints. Under some technical conditions, it is shown that the first proposed algorithm converges asymptotically to the desired equilibrium, while the second one converges to the solution in a practical way as long as the stepsize chosen is sufficiently small. Of particular interest is that the algorithms are designed to be robust to temporary errors in communication or computation. In addition, agents do not require global knowledge of total resources in the network or any specific procedure for initialization and the cost function is not required to be separable. The convergence of the algorithms is established via nonsmooth Lyapunov analysis. Finally, we illustrate the applicability of our strategies on a virus mitigation problem over computer and human networks.

A low-level resource allocation in an agent-based Cloud Computing platform

The distribution of computational resources in a Cloud Computing platform is a complex process with several parameters to consider such as the demand for services, available computational resources and service level agreements with end users. Currently, the state-of-the-art presents centralized approaches derived from previous technologies related to cluster of servers. These approaches allocate computational resources by means of the addition/removal of (physical/virtual) computational nodes. However, virtualization technology currently allows for research into new techniques, which makes it possible to allocate at a lower level. In other words, not only is it possible to add/remove nodes, but also to modify the resources of each virtual machine (low level resource allocation). Thus, agent theory is a key technology in this field, allowing decentralized resource allocation. This innovative approach has undeniable improvements such us computational load distribution and reduced computation time. The evaluation was carried out through experiments in a real Cloud environment, thus proving the validity of the proposed approach.

Decentralized Resource Allocation and Load Scheduling for Multicommodity Smart Energy Systems

Due to the expected growth in district heating systems in combination with the development of hybrid energy appliances such as heat pumps (HPs) and micro-combined heat and power (CHP) installations, new opportunities arise for the management of multicommodity energy systems, including electricity, heat, and gas. The possibility to convert forms of energy using hybrid energy appliances and exploiting flexibility from local production and consumption can improve the systems' efficiency significantly. This paper extends existing work with a decentralized version of a multicommodity smart energy management system to deal with flexibility and scalability. The system incorporates both heat and electricity, and integrates various types of flexible appliances as well as hybrid energy appliances. To optimally allocate the available resources and its flexibility, the developed multiagent system (MAS) aims to perform optimal supply and demand matching (SDM) of the local resources and flexible appliances, as well as to flatten out the net remaining exchange over time. The proposed method is applied to a test case, where simulation results confirm that the decentralized approach leads to a scalable solution for the management of the multicommodity smart energy system (MC-SES) and performs similar to the centralized approach.

Cost-Aware Coalitions for Collaborative Tracking in Resource-Constrained Camera Networks

We propose an approach to create camera coalitions in resource-constrained camera networks and demonstrate it for collaborative target tracking. We cast coalition formation as a decentralized resource allocation process where the best cameras among those viewing a target are assigned to a coalition based on marginal utility theory. A manager is dynamically selected to negotiate with cameras whether they will join the coalition and to coordinate the tracking task. This negotiation is based not only on the utility brought by each camera to the coalition, but also on the associated cost (i.e. additional processing and communication). Experimental results and comparisons using simulations and real data show that the proposed approach outperforms related state-of-the-art methods by improving tracking accuracy in cost-free settings. Moreover, under resource limitations, the proposed approach controls the tradeoff between accuracy and cost, and achieves energy savings with only a minor reduction in accuracy.

Polycentric structure and informal norms: competition and coordination within the scientific community

The success of the scientific community challenges in many ways our theories of social cooperation and public goods production. It is a very large-scale, decentralized, international organization lacking any central management or a formalized legislative or rule-enforcement body. Even the entry/exclusion rules are lax and unclear. By many standards it should not work. But, instead, it is one of the most successful human endeavors of all time. This paper provides an updated institutionalist theory of how this community works, with an extended discussion of its informal norms, prestige mechanisms, decentralized resource allocation, and interactions with states and civil society. Second, the paper discusses the ways in which the scientific community can fail at its truth-seeking task as a result of distortions created by outside political pressure and interactions with self-interested funding sources arguing that, as long as the polycentric structure is kept in place and the informal norms are preserved, the distortions are likely to be minor.

A balance of power principle for decentralized resource sharing

In a seminal 1978 paper Kleinrock (1979), Professor Leonard Kleinrock (“Kleinrock”) derived “rules of thumb” for flow-control, using a “power” metric of delay-throughput tradeoffs. This work stimulated a stream of, still on-going (e.g., Canzian et al., 2013), flow-control researches. A particularly challenging question, still widely open, has been: how can networks optimize a global performance index, such as power and through decentralized actions?At the same time, my Ph.D. thesis, under Len’s guidance, introduced microeconomic techniques for decentralized control of channel access schemes (Yemini and Kleinrock, 1979). This research too, continues to attract current interest (e.g., Chang et al., 2013).This paper introduces a generalized “power” metric for decentralized resource allocation, and uses it to derive novel Balance-of-Power principles, for Pareto optimal allocations. These principles substantially generalize and unify the results of Kleinrock (1979) and Yemini and Kleinrock (1979), as well as successor works, while shedding new light on resource sharing mechanisms in current virtual and cloud systems.

Organizational Decision-Making - Aggregating Individuals' Incentives and Information

How organizations - systems in which multiple actors with private information and heterogeneous incentives work towards a joint goal - reach decisions, is a question of fundamental importance to scholars from different domains in the management sciences. Notwithstanding the importance of prior work on group decision-making in the field of organizational behaviour, we believe that researchers' earlier single-sided focus on information aggregation within organizations stops short of explaining how agents with vested interests interact in the process of organizational decision-making. In this symposium, we seek to further the debate by presenting and discussing four papers that seek to address the above shortcoming. Searching in Teams: an Experimental Study of Aggregation and Incentives Presenter: Nils Stieglitz; Frankfurt School of Finance & Management Strategic Decision-Making in Hierarchies Presenter: Helge Klapper; U. of Vienna Presenter: Boris Maciejovsky; U. of California, Riverside Presenter: Phanish Puranam; INSEAD Presenter: Markus Reitzig; U. of Vienna Organizational Structure and Joint Learning Presenter: Michael Christensen; U. of Southern Denmark Presenter: Thorbjoern Knudsen; U. of Southern Denmark The Costs and Benefits of Decentralized Resource Allocation in Organizations: An Experimental Study Presenter: Reddi Kotha; Singapore Management U. Presenter: Jayanth Narayanan; National U. of Singapore Presenter: Phanish Puranam; INSEAD

Cooperative Decentralized Resource Allocation in Heterogeneous Wireless Access Medium

In this paper, radio resource allocation in a heterogeneous wireless access medium is investigated. Mobile terminals (MTs) are equipped with multiple radio interfaces and are assumed to have multi-homing capabilities. A novel algorithm, namely prediction based resource allocation algorithm, is proposed for the resource allocation. Unlike the existing solutions in literature, the proposed algorithm does not require a central resource manager to perform the radio resource allocation. The MT plays an active role in the resource allocation operation by requesting a bandwidth share from each available network based on the available resources at the network, such that the total allocated bandwidth from different networks satisfies the MT service requirement. The proposed algorithm is suitable for implementation in a dynamic environment with call arrivals and departures, and relies on network cooperation to perform the decentralized radio resource allocation in an efficient manner. Simulation results are presented to investigate the performance tradeoffs of the proposed algorithm.

Distributed Sum-Rate Maximization Over Finite Rate Coordination Links Affected by Random Failures

The deployment of small cell networks is expected to yield a significant improvement of coverage and spectral efficiency, provided that radio resource management is properly handled. In particular, in view of a potential massive deployment of femto access points (FAP's), it is advisable to consider decentralized resource allocation mechanisms able to keep interference below a suitable threshold. However, purely decentralized schemes might well be spectrally inefficient. In this paper, we consider a local coordination scheme, where nearby FAP's exploit the availability of a wired backhaul (typically an x-DSL line) to exchange control data to enable a local coordination with the aim of improving spectral efficiency. Since the backhaul is prone to random (unpredictable) delay and packet drop and the exchanged data are necessarily quantized, we propose a distributed resource allocation mechanism that limits the effects of the impairments coming from a realistic backhaul model. In particular, using results from stochastic approximation theory, we propose a distributed Robbins-Monro scheme with provable convergence properties. Numerical results are provided to validate the theoretical findings. Finally, in the case where the packet drop probabilities are known (estimated), we show how to counteract the effect of failures through a proper weighting of the messages exchanged among neighbor FAP's.

Local Public Good Provisioning in Networks: A Nash Implementation Mechanism

In this paper we study resource allocation in decentralized information local public good networks. A network is a local public good network if each user's actions directly affect the utility of an arbitrary subset of network users. We consider networks where each user knows only that part of the network that either affects or is affected by it. Furthermore, each user's utility and action space are its private information, and each user is a self utility maximizer. This network model is motivated by several applications including wireless communications. For this network model we formulate a decentralized resource allocation problem and develop a decentralized resource allocation mechanism (game form) that possesses the following properties: (i) All Nash equilibria of the game induced by the mechanism result in allocations that are optimal solutions of the corresponding centralized resource allocation problem (Nash implementation). (ii) All users voluntarily participate in the allocation process specified by the mechanism (individual rationality). (iii) The mechanism results in budget balance at all Nash equilibria and off equilibrium.

Iterative Spectrum Shaping with Opportunistic Multiuser Detection

This paper studies a multi-carrier based spectrum sharing system, in which two users iteratively update their respective transmit power allocation over parallel subcarriers (SCs) to maximize the individual transmit rate. Unlike the conventional iterative water-filling (IWF) algorithm that adopts the single-user detection (SD) at each user's receiver by treating the interference from all other users as additive noise, this paper proposes a new decentralized resource allocation scheme, namely iterative spectrum shaping (ISS), in which each receiver applies multiuser detection to opportunistically cancel the co-channel interference over selected SCs, thus termed opportunistic multiuser detection (OMD). Two coding schemes are investigated to exploit different forms of "frequency diversity" gains with OMD: carrier independent coding, which applies independent codebooks for different SCs to maximize the interference-decoding diversity; and carrier joint coding, which uses one single codebook across all the SCs to achieve the full coded diversity. For each coding scheme, the optimal transmit power allocation strategy is derived to maximize one user's transmit rate, with the power and rate allocation of the other user being fixed. Simulation results show that the converged system throughput by the proposed ISS algorithm with OMD is significantly improved over that by the conventional IWF with SD, in cognitive radio (CR) based wireless spectrum sharing networks.

Sensing time and power allocation for cognitive radios using distributed Q-learning

In cognitive radios systems, the sparse assigned frequency bands are opened to secondary users, provided that the aggregated interferences induced by the secondary transmitters on the primary receivers are negligible. Cognitive radios are established in two steps: the radios firstly sense the available frequency bands and secondly communicate using these bands. In this article, we propose two decentralized resource allocation Q-learning algorithms: the first one is used to share the sensing time among the cognitive radios in a way that maximize the throughputs of the radios. The second one is used to allocate the cognitive radio powers in a way that maximizes the signal on interference-plus-noise ratio (SINR) at the secondary receivers while meeting the primary protection constraint. Numerical results show the convergence of the proposed algorithms and allow the discussion of the exploration strategy, the choice of the cost function and the frequency of execution of each algorithm.

Optimal resource allocation for relay-assisted wireless communication systems

The design, analysis and optimization of cooperative/relaying communication systems have recently become a very active research area within both the information theory as well as communications engineering societies. It is now well understood that relaying strategies can improve the coverage of wireless networks by providing higher data rates or better transmission reliability to user terminals at the edge of a wireless cell, or terminals having faded connectivity with the base station. Relaying technologies are also becoming part of the telecommunication standards. Although we can find studies, in the academic literature, on advanced relaying schemes, which are based on user terminals cooperating to help each other while applying decentralized resource allocation strategies, the first actual deployment step which will take place within the 3GPP long-term evolution (LTE)-advanced standard is based on fixed access points to do the relaying and within a centralized scheme in which the e-nodeB (base station with backhaul connection) takes the scheduling and resource allocation decisions. One major objective in 3GPP evolution is to utilize the scarce wireless system resources efficiently because achieving the high quality of service (QoS) targets through over-provisioning is uneconomical due to the relatively high cost for transmission capacity in cellular access networks. Our objective here is to obtain the optimal (from an information theory perspective) resource allocation schemes taking into considerations the system constraints that are relevant to the LTE-advanced standard. We have been able to derive optimal resource allocation polices that are provably based on closed-form formulations which are practical for implementation. They include the policies for (i) transmission mode selection (i.e. deciding whether the user needs the assistance of a relay or not), (ii) power allocation for the base station and the relays, and (iii) criterion for user scheduling over the available air-link resource units. Simulation results demonstrate that our proposed resource allocation scheme provides considerable throughput gains especially for users receiving low power through the direct link with the base station.

Decentralized resource allocation to control an epidemic: A game theoretic approach

This paper examines how two countries would allocate resources at the onset of an epidemic when they seek to protect their own populations by minimizing the total number of infectives over the entire time horizon. We model this situation as a game between selfish countries, where players strategically allocate their resources in order to minimize the total number of infected individuals in their respective populations during the epidemic. We study this problem when the initial number of infectives is very small, which greatly simplifies the analysis. We show in this framework that selfish countries always allocate their resources so as to bring the effective reproduction ratio below one and avoid a major outbreak. When a major outbreak is avoidable, we further identify the necessary and sufficient conditions under which the individual allocation decisions of selfish countries match the decision that a central planner would make in order to minimize the total number of infectives in the whole population (without distinguishing between countries).

A Pricing Mechanism for Resource Management in Grid Computing

We consider the problem of efficient resource allocation in a grid computing environment. Grid computing is an emerging paradigm that allows the sharing of a large number of a heterogeneous set of resources. We propose an auction mechanism for decentralized resource allocation. The problem is modeled as a multistage stochastic programming problem. Convergence of the auction allocations to the social optimum is established. Numerical experiments illustrate the efficacy of the method.

Decentralized Resource Allocation in Dynamic Networks of Agents

We consider the problem of n agents that share m common resources. The objective is to derive an optimal allocation that maximizes a global objective expressed as a separable concave objective function. We propose a decentralized, asynchronous gradient-descent method that is suitable for implementation in the case where the communication between agents is described in terms of a dynamic network. This communication model accommodates situations such as mobile agents and communication failures. The method is shown to converge provided that the objective function has Lipschitz-continuous gradients. We further consider a randomized version of the same algorithm for the case where the objective function is nondifferentiable but has bounded subgradients. We show that both algorithms converge to near-optimal solutions and derive convergence rates in terms of the magnitude of the gradient of the objective function. We show how to accommodate nonnegativity constraints on the resources using the results derived. Experimental results with the problems of varying dimensions suggest that the algorithms are competitive with centralized approaches and scale well with problem size.

Revenue Generation and Its Consequences for Academic Capital, Values and Autonomy

The greatest challenge for institutions of higher education in most OECD countries since the 1970s has arguably been to cope with reduced public support. Many institutions responded to reductions in funding, first, by cutting costs and lobbying governments to reverse cutbacks, and then – when it became clear that funding levels would not be restored – by seeking out new sources of revenue. Some institutions decentralised resource allocation in order to encourage units to generate non-government revenue. Recent research into the revenue generation strategies of Canadian universities suggests, drawing upon the work of Pierre Bourdieu, that such measures, while potentially effective in stimulating resource acquisition – and beneficial in other important respects – change internal values and conditions in ways that may ultimately undermine universities’ autonomy, public credibility and capacity to create knowledge. Can leaders and managers enable their institutions to secure vital revenue, without diluting the values and conditions that have made universities unique and valuable to society? Can decision makers in government foster entrepreneurialism and responsiveness on the part of higher education institutions without compromising their raison d’être? This paper sheds light upon these questions.

A potential game approach to distributed power control and scheduling

Distributed solutions to resource allocation are motivated by the need to cope with the complexity in modern communication networks. The purpose of this paper is to discuss decentralized resource allocation in a self-organizing network from the viewpoint of potential games. The focus is on power allocation and scheduling in a congested distributed network such as a wireless ad hoc network. Noncooperative resource allocation is studied as a “potential” game, where the potential function is a common proxy objective, formalizing the implicit joint target of the noncooperative players. The potential function can be used to evaluate the system-level efficiency of noncooperative resource allocation. Examples of potential games are discussed in various contexts of distributed resource allocation. A game with discrete or convex strategy sets possessing a potential function has convergent greedy dynamics. A resource price determines the structure of a potential game.

Barriers to the Equitable Funding of Primary Healthcare in South Africa

Over 25 years after Alma-Ata, available evidence suggests that in low- and middle-income countries (LMIC) the funding and coverage of primary care is still inequitable. This article reviews the progress that has been made towards the equitable funding of primary healthcare (PHC) in South Africa and evaluates barriers to future progress. The South African experience is assessed to consider implications for other LMIC. The results show that substantial inequities in funding persist (albeit using a narrow definition of PHC). Underlying causes relate to fiscal austerity, the lack of protection of PHC funding, incremental resource allocation and the belief that poorer districts are not able to use extra funds effectively. These results match the experience in other LMIC. Central governments have a critical role to play in lobbying for and protecting resources for PHC, maintaining more control over decentralised resource allocation and building capacity to benefit in poorer areas.

Distributed Scheduling and Dynamic Pricing in a Communication Network

This paper studies dynamic resource allocation in a decentralized communication network. The temporal aspect in the decentralized resource allocation problem presents new challenges, e.g., in optimizing the delay-throughput trade-off under user-specific delay costs. A dynamic bandwidth allocation game modelling an agent-based network is presented. The dynamic noncooperative game achieves Pareto-efficient bandwidth allocation that can be implemented by a greedy algorithm with pricing. Optimal dynamic pricing is discussed for the efficient sharing of network resources. An ad hoc wireless network is an example of such self-organizing decentralized system: the mobile nodes need not be directly connected to a base station. Another application of the model is to consider distributed uplink scheduling, based on local information, in a WCDMA network. The discretized control variable of a mobile node is either the received power/QoS-level or the binary decision on packet transmission.