Search reduction in hierarchical distributed problem solving

Knoblock and Korf have determined that abstraction can reduce search at a single agent from exponential to linear complexity (Knoblock 1991; Korf 1987). We extend their results by showing how concurrent problem solving among multiple agents using abstraction can further reduce search to logarithmic complexity. We empirically validate our formal analysis by showing that it correctly predicts performance for the Towers of Hanoi problem (which meets all of the assumptions of the analysis). Furthermore, a powerful form of abstraction for large multiagent systems is to group agents into teams, and teams of agents into larger teams, to form an organizational pyramid. We apply our analysis to such an organization of agents and demonstrate the results in a delivery task domain. Our predictions about abstraction's benefits can also be met in this more realistic domain, even though assumptions made in our analysis are violated. Our analytical results thus hold the promise for explaining in general terms many experimental observations made in specific distributed AI systems, and we demonstrate this ability with examples from prior research.This research has been sponsored, in part, by the National Science Foundation under grants IRI-9015423 and IRI-9010645, by the University of Michigan Rackham Graduate School, and by a Bell Northern Research Postgraduate Award.     

A Formal Study of Distributed Meeting Scheduling

Automating routine organizational tasks, such as meeting scheduling, requires a careful balance between the individual (respecting his or her privacy and personal preferences) and the organization (making efficient use of time and other resources). We argue that meeting scheduling is an inherently distributed process, and that negotiating over meetings can be viewed as a distributed search process. Keeping the process tractable requires introducing heuristics to guide distributed schedulers' decisions about what information to exchange and whether or not to propose the same tentative time for several meetings. While we have intuitions about how such heuristics could affect scheduling performance and efficiency, verifying these intuitions requires a more formal model of the meeting schedule problem and process. We present our preliminary work toward this goal, as well as experimental results that validate some of the predictions of our formal model. We also investigate scheduling in overconstrained situations, namely, scheduling of high priority meetings at short notice, which requires cancellation and rescheduling of previously scheduled meetings. Our model provides a springboard into deeper investigations of important issues in distributed artificial intelligence as well, and we outline our ongoing work in this direction.     

The Effects of Electronic Collaboration in Distributed Project Management

Abstact The emergence and widespread use of collaborative technologies for distributed project management has opened up opportunities for off-shore outsourcing and collaborative development. However, most tools and techniques for project management focus on on-site, long term relationships and sourcing strategies at a time when inter-organizational relationships are becoming more dynamic and geographically dispersed. This paper uses grounded theory to analyze data on virtual teams' projects. The analysis uncovers “effects” in the way in which distributed projects are managed. These effects relate to coordination, communication and adaptation to distributed electronic work environments. Following an analysis of these electronic collaboration “effects”, a model for distributed project management is presented.     

A Time-Bound Framework for Negotiation and Decision Making of Virtual Manufacturing Enterprise

Virtual manufacturing has 2 characteristics as an agent-based electronic commerce environment: dynamic nature of resource status and variety of agents' decision-making (i.e., scheduling) model. To reflect the characteristics, a relevant negotiation protocol should be designed and an appropriate decision-making model should be developed. In this article, from the perspective of a sales agent that is a middle man between customers and manufacturers in a virtual manufacturing environment, we provide a case study that suggests a time-bound framework for external negotiation between sales agents and customer agents, and internal cooperation between sales agents and manufacturing agents. We assume a job shop as the production model of a virtual manufacturing enterprise and formulate the optimal order selection problem with mixed integer programming, but its computation time is not acceptable for real-world problems. For this time-constrained decision making, we develop a genetic algorithm as an anytime problem-solving method for the scheduling of the production model, which shows a reasonable computation time for real-world cases and good incremental problem-solving capability.     

关于分散系统管理的移动模式的设计

本文认为,随着通讯技术的迅速发展,对于大规模分散系统管理的要求越来越高,因为这样的系统组成要素是不均匀或是异质性的,而且各个组成要素都会形成不同的信息,管理时会生成大量的项目输送量。为了有效地管理大规模分散系统,需要对发生项目进行过滤,并相机变换管理系统,增强其扩张性和减轻网络系统的负担。为此而进行的在分散系统基础上使用移动agent的研究,设计能够有效地移动agent的移动模式,协调以前移动agent的移动方法与应答提案方法之间的关系。     

An Architecture for Conflict Management Between Make and Delivery Agents in an Enterprise Supply Chain

ABSTRACT

Problems related to conflict management resolution between Make and Delivery activities within a Supply Chain are studied here. The traditional approach, which usually manages the problem via a lobal optimization system, seems not suitable due to the fact that Make and Delivery may belong to different enterprises. Furthermore, the prevalence of dynamic scheduling concepts is preferred to the rigidity of a well-fixed optimized plan. Every agent within this process has a traditional production/delivery plan, but communication is performed in terms of a generic common plan that states the dates of reception/delivery. The software architecture we propose uses a distributed and fully decentralized agent model; the customization of such a system has been accomplished by adding a congruent number of business processes, which manage plan synchronization problems, using interchangeable strategies. A dedicated agent that retrieves all the information useful to infer statistical evaluations has been included to gain feedback on system performance.     

A‐pool: An agent‐oriented open system shell for distributed decision process modeling

An agent‐oriented open system shell, A‐Pool, for distributed decision process modeling in the Internet domain is presented. Unlike most decision support systems, A‐Pool provides a testbed for modeling and understanding the cognitive aspects of distributed decision processes themselves rather than for domain‐specific problem solving. This is achieved with a pool of virtual agents and a pool of cognitive maps of the agents at each A‐Pool node. The virtual agent scheme extends object‐oriented programming to the Internet domain and supports different communication and collaboration protocols with virtual communities, virtual sessions, and virtual conferences. The cognitive map scheme supports perspective sharing and various conflict integration and resolution strategies through cognitive map composition, derivation, and focus generation. Thus each A‐Pool node provides an architecture for modeling interdependencies and for ensuring global coherence; in addition, the communication is asynchronous and the control is distributed, allowing a large degree of autonomy and the examination of various thoughts and social protocols involved in strategic planning in an open system environment. Basic ideas are illustrated with a running example.     

Meta-level Coordination for Solving Distributed Negotiation Chains in Semi-cooperative Multi-agent Systems

A negotiation chain is formed when multiple related negotiations are spread over multiple agents. In order to appropriately order and structure the negotiations occurring in the chain so as to optimize the expected utility, we present an extension to a single-agent concurrent negotiation framework. This work is aimed at semi-cooperative multi-agent systems, where each agent has its own goals and works to maximize its local utility; however, the performance of each individual agent is tightly related to other agents’ cooperation and the system’s overall performance. We introduce a pre-negotiation phase that allows agents to transfer meta-level information. Using this information, the agent can improve the accuracy of its local model about how other agents would react to the negotiations. This more accurate model helps the agent in choosing a better negotiation solution for a distributed negotiation chain problem. The agent can also use this information to allocate appropriate time for each negotiation, hence to find a good ordering of all related negotiations. The experimental data show that these mechanisms improve the agents’ and the system’s overall performance significantly.     

静态分簇下的大规模分布式天线系统用户调度算法

在大规模分布式天线系统中,静态分簇和用户调度带来的簇间干扰问题会导致系统和速率下降。针对这个问题,提出了一种两阶段贪婪用户调度算法。首先,每个簇内并行实施贪婪用户调度;然后,从全局上再次利用贪婪算法来剔除簇间干扰较大、服务质量较差的用户,使得系统和速率进一步提升。仿真结果表明,随着不同系统参数的改变,两阶段贪婪用户调度算法可有效提高系统和速率。     

Coordination Structures and System Restrictiveness in Distributed Group Support Systems

Most GSS research has studied the impact of restricting group interaction to GSS-prescribed coordination structures with face-to-face groups, while Distributed GSS (DGSS) has been largely ignored. Due to the nature of mediated communication in asynchronous interaction, it is relatively difficult to coordinate distributed groups, and a special coordination structure must be arranged to overcome these difficulties. This study examines the effect of system restrictiveness of coordination structures in an asynchronous environment. A 2 × 2 factorial experiment was designed with two independent variables – sequential vs. parallel coordination mode, and with vs. without a leader – to construct coordination structures with varying degrees of restrictiveness. The study finds that less restrictive coordination structures are more appropriate to support asynchronously interacting distributed groups. Objective decision quality is equal for both parallel and sequential coordination mode, but is significantly better with a group leader. Groups with parallel coordination mode have a stronger belief that the decisions they made are of higher quality than those of groups with sequential coordination mode. In groups with a leader, communication effectiveness is better. Satisfaction with a decision process is higher in parallel coordination groups and in groups with a leader. There is also a significant interaction effect. Satisfaction with the decision process is higher in sequential coordination groups with a leader than sequential coordination groups without a leader.     

Inefficiencies in a Model of Team Formation

We consider a two period model with complete information involving three agents, two on one side of the market, and one on the other. The agents on the same side of the market bargain, among themselves, over whether to form a team or not, and also with the other agent, either singly or as a team, regarding the payoffs. We characterize the subgame perfect Nash equilibrium of this game. We find that the behavioral assumptions regarding the agents play a critical role in the outcome. If the agents are combative then the outcome is efficient. Whereas if the agents are peace loving, and the discount factor is large, then the outcome may involve delay, as well as (inefficient) team formation.     

Intelligent Cyber Logistics Using Reverse Auction in Electronic Commerce

In electronic markets, customers can purchase products online. In addition, the seller or an Internet shopping mall has to deliver the products to the buyers. Transportation agencies can be chosen to deliver the products from an Internet shopping mall. Much research about efficient and automatic delivery scheduling has been carried out on delivery-ordering. By and large, transportation agencies need to maintain not only a cooperative relationship but also a competitive relationship among themselves to meet with customers' preferences for their purchases. This article suggests an intelligent, cyber-logistic framework as a multiagent system and presents designs for an intelligent software-agent structure to maintain this cooperative and competitive relationship. This intelligent cyber-logistic system consists of shopping mall agents, a broker agent, and delivery and scheduling agents. These components, based on the concept of reverse auction, lead to improvement in the overall system efficiency as well as increased profits for each transportation company. Traditionally, in researching vehicle routing problems (VRP), the cooperative and competitive relationships among the transportation companies have not been addressed. However, in this article, we have addressed these issues. We also suggest the reverse method of auction, which solves VRP for their own trucks to provide the transportation of goods for the shopping malls. After finding their own solution, the transportation company can bid with its own cost of delivery. We have used the constraint satisfaction problems solving method to conduct truck scheduling for each transportation agency. To perform the delivery-ordering process between agents, we have defined the logisticsKQML to represent the message protocol of upper level. Finally, we have evaluated the cyber-logistic system using reverse auction by a simulation experiment and have discussed further research and drawn conclusions.     

The Precursor Role of Cooperation,Coordination, and Relationship Assets in a Relationship Model

ABSTRACT

Purpose: The primary purpose of this study is to test the importance of activity-oriented precursors in a relationship model. This study supports the theoretical view that firms create trust and knowledge through activities and these activities make a commitment decision less risky (Johanson & Vahlne, 2009). The secondary purpose of this study is to collect and examine data from interorganizational relationships in both Sweden and the United States. By including data from two countries, results will be more generalizable. Results can also lead to several managerial implications.

Methodology/approach: This study focuses on a sample of distributors from both the United States and Sweden. One hundred sixty-one usable surveys were returned from the U.S. survey, for a response rate of 27%. One hundred twenty-four usable surveys were returned from the Swedish survey, for a response rate of 21%. The PLS-SEM method was used to examine the model’s constructs.

Findings: Similar to past research results show that trust and commitment have a direct positive influence on satisfaction, and that trust also has a direct positive influence on commitment. However, this study uniquely supports four out of six newly tested hypotheses. Both cooperation and relationship assets have a direct positive influence on commitment. Cooperation has a direct positive influence on trust and commitment. Relationship assets have a direct negative influence on trust but a direct positive influence on commitment. Surprisingly, two hypotheses were not supported: Coordination did not have a significant relationship with either trust or commitment.

Research implications: Managers who want to achieve a satisfactory relationship based on trust and commitment need to prioritize their attention toward cooperation. They should also be aware that participation in joint activities (i.e., coordination and relationship investments) does not guarantee higher levels of trust or commitment in the relationship. It is the quality of the joint activities and the how dependent firms are on each other and not just participation in joint activities that are likely to create higher levels of trust or commitment. The quality of coordination and manageable levels of dependence may counteract the higher costs associated with joint activities compared to the costs associated with cooperation. Managers may be wise to not make major commitments to other firms unless high quality joint activities have created knowledge and trust between firms. Originality/value/contribution: The model adds the joint activity-oriented antecedents associated with collaboration which is essential to a successful relationship. Because of the high failure rate of collaboration may be due to cooperation and coordination failures and because these two constructs are underspecified in interorganizational research, this study is unique in examining activity-oriented antecedents in a trust/commitment model of relationship satisfaction in a cross-cultural context (i.e., with U.S. and Swedish samples).     

自适应选择的雷达回波空间积累检测方法

空间部分相干时的回波积累检测是分布式雷达回波信号联合处理面临的难题。针对传统相干和非相干积累检测方法在空间部分相干时回波检测性能损失较大的问题,建立了目标三维散射模型,分析了各站回波空间相关性变化规律,引入时域加权能量检测器用于空间部分相干积累检测,提出了以回波的空间相关系数大小为依据进行自适应选择的积累检测方法,有效解决了回波在任意空间相关性时高效积累检测。仿真结果表明该方法能根据相关系数大小正确选择3种积累检测方法,且检测性能优于传统的相干和非相干积累检测方法。     

Dynamic Agency and Investment Theory under Model Uncertainty

We extend dynamic agency and investment theory by incorporating model uncertainty. As concerns regarding model uncertainty induce a trade‐off between incentives and ambiguity sharing, the principal tends to delay the cash payout to the agent. We find model uncertainty lowers the firm value, the average q and marginal q, where q is defined as the ratio between a physical asset's market value and its replacement value. Furthermore, model uncertainty leads to insufficient investment, which provides an alternative explanation for under‐investment. Finally, the optimal pay‐performance sensitivity of the agent's continuation value to the firm's output is state dependent and exceeds the lower bound when it is close to the payout boundary.     

基于列生成的毫米波多连接链路配置调度算法

多连接技术允许用户同时建立和保持与多个小区/接入点的连接，通过网络元素之间的协调在吞吐量和可靠性方面大幅提高网络性能。针对毫米波通信中超高频段的链路中断问题，研究了多连接基于链路配置的调度算法，以提高链路调度效率，降低复杂度。首先，在系统模型中采用链路配置作为优化变量；其次，设计了多连接比例公平的调度准则；最后，提出一种基于列生成算法的链路配置调度优化算法，利用Dantzig-Wolfe分解将原问题分解为限制主问题和定价问题，并结合分支定界方法获得最优解。仿真结果表明，所提算法能够在数值上逼近全局最优，并且比现有的毫米波蜂窝网络链路调度方案增益平均提高40%以上。     

The role of channel quality in customer equity management

Maximization of customer equity is a core objective of customer–company relationship management. We present an extended model of customer equity for determining the optimal allocation of marketing resources across acquisition and retention activities. Focusing on the negative relationship between acquisition and retention, we motivate channel quality as a relevant decision variable, explicate its role in the model, and demonstrate the existence of an optimal value. In addition, rather than making concavity assumptions about acquisition and retention rate response curves, we use the flexible ADBUDG model (Little, JDC, Models and Managers: the Concept of a Decision Calculus. Manag Sci 1970; 16(8): 466–484.), which allows for both S-shaped and strictly-concave relationships, and parameterize it using decision calculus. We show how to estimate and apply the model and then provide sensitivity analyses with respect to changes in the true values of model parameters as well as inaccuracy in managerial inputs. We conclude by comparing our model with extant models and discussing the implications of our research.     

Information Asymmetries as Antecedents of Opportunism in Buyer-Supplier Relationships: Testing Principal-Agent Theory

ABSTRACT

Purpose: A severe problem in supplier selection refers to moral hazard: suppliers not behaving in the expected way once contracted. Principal-agent theory could provide insights on how to reduce this problem. Because buyer–supplier relationships can be interpreted as principal-agent situations, the application of agency theory should facilitate improved supplier selection. Although theoretically compelling, empirical tests verifying this assumption are not prevalent. Regarding the advancement of theory, this paper tests whether both ex ante and ex post information asymmetries influence moral hazard. In particular, in the context of a globalizing economy with a subsequent increase in information asymmetries as a problem in supplier selection, this conceptual approach may be contributive.

Design/methodology/approach: The authors use a set of 87 buyer–supplier relationships to conduct a test, applying a partial least squares model with latent variables. A particularity of the data set is that it contains information on ongoing as well as on discontinued relationships.

Findings: The analysis indicated that both ex ante information asymmetries (operationalized by a reputation variable) and ex post asymmetries (operationalized by a monitoring variable) have shown to be significant and strong antecedents explaining the occurrence of moral hazard. Interestingly, and opposed to the common assumption, the length of a relationship and the amount of direct meetings have not revealed any explanatory significance. Buyer dependency hardly showed influence on supplier opportunism.

Research limitations/implications: Data were collected from a multitude of buyer–supplier relationships from a single firm in the chemical-pharmaceutical industry. Generalizations to other industries still need to be tested. Socially desirable answering behavior cannot fully be excluded because relationship discontinuation is not a desirable situation. In terms of theory implications, this research adds to the notion that both hidden action and hidden intention can lead to moral hazard.

Practical implications: An agency-based analysis can be operationalized with the help of an agency-based supplier classification portfolio. It might be of particular value to firms to discuss those suppliers that scored high in risk of opportunism but did not (yet) reveal any signs of moral hazard. Finally, the strong explanatory power

of reputation alerts buyers to pay more attention to behavioral information on the (potential) supplier available in the market.

Originality/value: Analyzing the occurrence of moral hazard and including terminated relationships adds to the emerging stream of literature on relationship discontinuation in B2B markets. Further, the strong empirical results may encourage researchers to elaborate on principal-agent theory-based assumptions, adding another layer of explanation to buyer–supplier relationships. Findings show that reputation is unduly neglected as supplier selection criterion in current theory and practice.     

An Intelligent Distributed System for Strategic Decision Making

The decision-making process in strategic planning is often too complex to be handled by conventional methods. Strategic planning problems (building new plans, new product planning, etc.) belong to the class of problems called ill-structured by H. Simon. They involve a decomposition of the main problem into a set of subproblems, a reasoning process at the subproblem level, and then a coordinated and aggregated process to build a global solution. Because partial solutions are generated without having a complete view of the global objective, this type of decision-making process very often generates incoherent and contradictory hypotheses and actions. Therefore, the main problem is to find a way to achieve coherence and coordination among decisions made locally by different agents, at different levels.Recent advances in Artificial Intelligence, particularly in the field of multi-agent theory, offer great promises in modeling strategic planning processes. In this article we present a general framework called 'A Coherent Plan of Coordinated Actions (CPCA)' for building intelligent distributed strategic decision making systems which integrates advances in both distributed decision making and Distributed Artificial Intelligence. We then describe a multi-blackboard system, ARISTOTE, which is aimed at helping corporate managers address the feasibility and coherence of a plan of actions.