Information Technology Reference
In-Depth Information
can be cooperative with each other for multi-tasks performed in a plant. They proposed a modeling
framework to ease the design of autonomous agents for large-scale production systems. Different from
the applications of cooperative multi-agents in other industries, the focus of our work and contribu-
tion is to develop a trading cooperative multi-agent system to increase trading efficiencies in matching
volumes in an auction environment, at a competitive price.
Collaboration between buyer agents and seller agents, respectively, is the basis of the mechanism
of matching trading parameters to close a trade where no trade would be possible because of the mis-
match between the buyer's and seller's trading volume. Each buyer and each seller desire to achieve a
trade consistent with the quantity of the buyer's bid, or the seller's offer. There is a cost associated with
non-completion, necessitating a carry-over to the next period. This chapter presents the impact of con-
sortiums formed by the cooperation between buyer agents, and seller agents, respectively, to increase
volume by combining bids and offers, as needed, as a means of matching and closing a trade. Thus,
trading volumes may be increased, with the associated reduction of trading inventory and carry-overs
of trading volumes to succeeding periods.
The remainder of the chapter is organized as follows. A mathematical programming model using
dynamic programming is described to model the decision problem of volume matching by consortium.
An analytical solution to the dynamic programming model is difficult to achieve in real time, and is
impractical in a real world auction environment. To circumvent this situation, heuristic rules of model-
ing to quickly analyze and solve the problem are developed and presented. Also shown in this section
are experimental results of auction trades by consortium from randomly generated trade offers. The
conclusions of the study are stated, together with ideas and directions for future research.
Model for agent t ransactions
We consider a two-tier supply chain composed of
m
suppliers and
n
buyers with
k
types of products.
The two tiers consist of sellers and buyers. Buyers buy directly from sellers, with no intermediaries,
negotiators, or other traders. It is a simple construct of suppliers and buyers, respectively, bidding for
the goods directly to one another. The construct has been kept simple to emphasize the complexity of
the decision process and trading actions, and to focus on the best practices of decision-making in this
trading environment.
An agent makes decisions and performs actions in trading in a difficult environment of uncertainty
where nothing is known about the future. A seller agent has no information about the future needs of
buyer agents,
i.e.
about the quantities that might be asked for and the future possibility of a better match
with the seller agent's offer. A buyer agent has no information about future offers by a seller agent,
and the future possibility of a better match of the buyer agent's quantity requirement with the quantity
offered by a future seller agent's offer. The only information available is the present set of conditions.
These are described as follows:
1. A buyer agent is aware of the selling bids placed by seller agents.
2. A seller agent is aware of the buying bids placed by buyer agents.
3. A buyer agent is aware of the community of buyer agents and the quantities requested by each
buyer agent.
Search WWH ::
Custom Search