A mathematical framework for asynchronous, distributed, decision-making systems with semi-autonomous entities: algorithm synthesis, simulation, and evaluation

Lee, T.S.; Ghosh, Sumit; Nerode, Anil

doi:10.1109/isads.1999.838435

Cited by 7 publications

(7 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…• Reference model: simulates the dynamic representation of the enterprise process under control (see Kohn et al [1994], Lee at al [1994], Nerode et al [1992], Nerode et al [1993]). • Control Generator: computes the optimal control law as a causal map from the performance space ( the product of the sensor space , the parameter space the model frame space and the state space) to the space of actions.…”

Section: Fig 2 Conceptual Enterprise Control Architecturementioning

confidence: 99%

Repair-Control of Enterprise Systems Using RFID Sensory Data

Kohn

Brayman²

2008

IFAC Proceedings Volumes

View full text Add to dashboard Cite

Section: Fig 2 Conceptual Enterprise Control Architecturementioning

confidence: 99%

Repair-Control of Enterprise Systems Using RFID Sensory Data

Kohn

Brayman²

2008

IFAC Proceedings Volumes

View full text Add to dashboard Cite

“…The most direct approach t o s u c h implementations is to directly parallelize AI production systems or the underlying programming languages 93,221]. An alternative and more challenging approach is to use distributed computing, where not only are the individual reasoning, planning and scheduling AI tasks parallelized, but there are di erent modules with di erent s u c h tasks, concurrently working toward a common goal 137,138,166].…”

Section: Distributed Arti Cial Intelligencementioning

confidence: 99%

Theory of Collective Intelligence

Wolpert

2004

Collectives and the Design of Complex Systems

View full text Add to dashboard Cite

This paper surveys the emerging science of how to design a \COllective INtelligence" (COIN). A COIN is a large multi-agent system where: i) There is little to no centralized, personalized communication and/or control. ii) There is a provided world utility function that rates the possible histories of the full system.In particular, we a r e i n terested in COINs in which e a c h a g e n t runs a reinforcement learning (RL) algorithm. The conventional approach to designing large distributed systems to optimize a world utility does not use agents running RL algorithms. Rather, that approach begins with explicit modeling of the dynamics of the overall system, followed by detailed hand-tuning of the interactions between the components to ensure that they \cooperate" as far as the world utility i s concerned. This approach is labor-intensive, often results in highly nonrobust systems, and usually results in design techniques that have limited applicability. In contrast, we w i s h t o s o l v e the COIN design problem implicitly, via the \adaptive" character of the RL algorithms of each of the agents. This approach i n troduces an entirely new, profound design problem: Assuming the RL algorithms are able to achieve high rewards, what reward functions for the individual agents will, when pursued by those agents, result in high world utility? In other words, what reward functions will best ensure that we d o n o t have phenomena like t h e tragedy of the commons, Braess's paradox, or the liquidity trap? Although still very young, research speci cally concentrating on the COIN design problem has already resulted in successes in arti cial domains, in particular in packet-routing, the leader-follower problem, and in variants of Arthur's El Farol bar problem. It is expected that as it matures and draws upon other disciplines related to COINs, this research will greatly expand the range of tasks addressable by human engineers. Moreover, in addition to drawing on them, such a fully developed science of COIN design may provide much insight into other already established scienti c elds, such as economics, game theory, and population biology.

show abstract

“…The potential benefits of ADDM systems include scalability, high throughput, and robustness. Examples of ADDM systems include the distributed routing algorithm for ATM networks [22], YADDES [23], NO-VAHID [24], DICAF [25], DARYN [26], RYNSORD [27], MFAD [28], and ATOS [29]. The ATOS: Autonomous Decentralized Transport Operation System, controls the world's largest transportation system-East Japan Railway Company.…”

Section: Formal Definition Of Stability In Addm Systemsmentioning

confidence: 99%

“…"A" and "B" constitute the two basic manifestations of ADDM systems. Examples of "A" include RYNSORD [27] and NOVAHID [24] while that of "B" include MFAD [28]. ADDM systems, by nature, are large-scale and complex.…”

Section: Formal Definition Of Stability In Addm Systemsmentioning

confidence: 99%

“…3.1 Overview of the MFAD System MFAD [28] refers to a mathematical framework, based on the Kohn-Nerode [32] distributed hybrid control paradigm, wherein a decentralized military command and control decision-making algorithm is synthesized from a centralized description of the problem. In contrast to the traditional centralized control, the decentralized model is increasingly appropriate in today's military, given the sophisticated training, education, and data gathering, fusing, and processing capability at the level of every individual unit.…”

Section: Stability Analysis Of a Decentralized Military Command And Cmentioning

confidence: 99%

See 1 more Smart Citation

The concept of "stability" in asynchronous distributed decision-making systems

Lee

Ghosh

2000

IEEE Trans. Syst., Man, Cybern. B

Self Cite

View full text Add to dashboard Cite

Asynchronous distributed decision-making (ADDM) systems constitute a special class of distributed problems and are characterized as large, complex systems wherein the principal elements are the geographically dispersed entities that communicate among themselves, asynchronously, through message passing and are permitted autonomy in local decision making. Such systems generally offer significant advantages over the traditional, centralized algorithms in the form of concurrency, scalability, high throughput, efficiency, low vulnerability to catastrophic failures, and robustness. A fundamental property of ADDM systems is stability that refers to their behavior under representative perturbations to their operating environments, given that such systems are intended to be real, complex, and to some extent, mission-critical, and are subject to unexpected changes in their operating conditions. This paper introduces the concept of stability in ADDM systems and proposes an intuitive yet practical and usable definition that is inspired by those used in control systems and physics. An ADDM system is defined as a stable system if it returns to a steady state in finite time, following perturbation, provided that it is initiated in a steady state. Equilibrium or steady state is defined through placing bounds on the measured error in the system. Where the final steady state is equivalent to the initial one, a system is referred to as strongly stable. If the final steady state is potentially worse then the initial one, a system is deemed marginally stable. When a system fails to return to steady state following the perturbation, it is unstable. The perturbations are classified as either changes in the input pattern or changes in one or more environmental characteristics of the system, such as hardware failures. For a given ADDM system, the definitions are based on the performance indices that must be judiciously identified by the system architect and are likely to be unique. To facilitate the understanding of stability in representative real-world systems, this paper reports the analysis of two basic manifestations of ADDM systems that have been reported in the literature: (1) a decentralized military command and control problem, MFAD and (2) a novel distributed algorithm with soft reservation for efficient scheduling and congestion mitigation in railway networks, RYNSORD. Stability analysis of MFAD and RYNSORD yields key stable and unstable conditions. A system determined to be stable provides the reassurance that the system will perform well under adverse conditions. In contrast, a system deemed unstable reflects the need to address key weaknesses in the system design. Thus, stability analysis is a necessary and critical step in the development of any ADDM system.

show abstract

A mathematical framework for asynchronous, distributed, decision-making systems with semi-autonomous entities: algorithm synthesis, simulation, and evaluation

Cited by 7 publications

References 12 publications

Repair-Control of Enterprise Systems Using RFID Sensory Data

Repair-Control of Enterprise Systems Using RFID Sensory Data

Theory of Collective Intelligence

The concept of "stability" in asynchronous distributed decision-making systems

Contact Info

Product

Resources

About