Providing efficient mechanism for allocation and management of resources is essential for realizing a computational grid. Since resource providers and consumers may be independent bodies in a computing grid platform, negotiation among these participants is
47. K. M. Sim
by trading conventions, or strategies encoded by users. Although Kasbah considered deadline as a factor when selecting the strategy for making concession before trading commences, such selection cannot take into account the ever-changing external influences such as increasing/decreasing number of producers and consumers. Using negotiation decision functions (NDFs) based on time, resource, and the behavior (attitude) of its opponents, Faratin et al. [18] defined a range of tactics for generating (counter-)proposals. While time-NDFs consider deadlines, resource-NDFs take into consideration diminishing level of resources when determining the amount of concession. Behavior-NDFs allow agents in [18] to replicate the attitude of their trading partners. Like Kasbah, Faratin et al.’s NDFs do not address design issues (i) and (iii) mentioned above. FeNAs[19] adopts fuzzy constraint-based reasoning for bolstering multi-issue negotiations. One of the distinguishing features of FeNAs is that preferences, priorities and constraints are defined as fuzzy constraints (e.g. low price, high quality, short delivery time and budget at about $90). FeNAs’s objectives may also be defined as soft objectives (e.g. an agent prefers to pay $200 but is still happy with paying a little bit more). Although this feature may provide FeNA with the flexibility of acquiring available suboptimal resources more quickly, it does not take into account market factors as well as time constraints. In a highly dynamic grid environment, it is essential to take market dynamics into consideration because (i) providers can make resources/services available to and disconnect from a market-oriented grid, and (ii) consumers can enter and withdraw requests, perhaps at machine speed in both cases. On this account, it seems prudent to design negotiation agents in a market-oriented grid by considering market factors (such as opportunity and competition) that can significantly enhance or diminish the successful acquisitions (or provisions) of resources/services with optimum returns.
3. MARKET-DRIVEN AGENT
Sim et al. [10-15] proposed several variants of market-driven strategies for negotiation agents that make concessions taking into account factors such as outside options, market rivalry, and time pressure. A market-driven agent (MDA) is a negotiation agent that makes adjustable amounts of concession by considering deadline, competition, and opportunity. Due to space limitation, this section only presents a very brief summary of the features of MDAs, and detailed designs and previous empirical results evaluating MDAs can be found in [10-15]. An MDA makes concession by narrowing the difference kt between its proposal and the counter-proposal of its opponent in a negotiation round t. In determining the appropriate amount of concession, an MDA adopts the following three decision functions to determine the difference kt+1 in proposal and counter-proposal in the next round t+1:
kt+1= f[O(nBt, vB→Sjt, <wSj→Bt>),C(mBt,nBt),T(t,τ,λ)]kt
The above formulation models the decision functions from a buyer agent’s perspective (the formulation for a seller agent is similar). Let B be a buyer agent. The opportunity function O(nBt, vB→Sjt, <wSj→Bt>) of B determines the amount of concession based on (i) nBt –which is the number of trading alternatives of B (i.e., the number of sellers) and (ii) differences in utilities (vB→Sjt) generated by the proposal of B and the counter-proposal of its trading partner(s) (<wSj→Bt>) [11,12]. vB→Sjt is the utility that B will receive if seller Sj accepts B’s proposal. <wSj→Bt> is a set of utilities such that each wSj→Bt is the utility that B will receive if B accepts seller Sj’s proposal. When determining opportunity, it was shown in [11] that if there is a large number of trading alternatives, the likelihood that ACM SIGecom Exchanges, Vol. 5, No. 2, November 2004, Pages 45–53.
(Market-oriented Process Organization of R&D) R&D 研发变革管理(Change ...(Market-Driven NPD Process Management) 研发项目管理(R&D Project Management)...
Market-driven energy pricing necessary to ensure China...It will be argued that a market-oriented ...At the same time, investment in grids annually ...
ON GRIDS AND SOLUTIONS FROM RESIDUAL MINIMIZATION by Hiroaki Nishikawa A dissertation...A little surprisingly, the grid movement driven by this simple minimizat...
A Market Mechanism for Energy Allocation in Micro-CHP Grids 微网能量的市场...A study from Emerging Energy Research [16] states that today wind power ...
热门文档