Agent based interactions and economic

Agent-based Interactions and Economic
Encounters in an Intelligent Inter Cloud
Abstract
An InterCloud is an interconnected global “Cloud of Clouds” that enables each Cloud to
tap into resources of other Clouds. This is the earliest work to devise an agent-based
InterCloud economic model for analyzing consumer-to-Cloud and Cloud-to-Cloud
interactions. While economic encounters between consumers and Cloud providers are
modeled as a many-to-many negotiation, economic encounters among Clouds are
modeled as a coalition game. To bolster many-to-many consumer-to-Cloud negotiations,
this work devises a novel interaction protocol and a novel negotiation strategy that is
characterized by both 1) adaptive concession rate (ACR) and 2) minimally sufficient
concession (MSC). Mathematical proofs show that agents adopting the ACR-MSC
strategy negotiate optimally because they make minimum amounts of concession. By
automatically controlling concession rates, empirical results show that the ACR-MSC
strategy is efficient because it achieves significantly higher utilities than the fixed-
concession-rate time-dependent strategy. To facilitate the formation of InterCloud
coalitions, this work devises a novel four-stage Cloud-to-Cloud interaction protocol and a
set of novel strategies for InterCloud agents. Mathematical proofs show that these
InterCloud coalition formation strategies 1) converge to a subgame perfect equilibrium
and 2) result in every Cloud agent in an InterCloud coalition receiving a payoff that is
equal to its Shapley value.
Existing System:
In Many existing research they only consider the power consumption cost. As a
major difference between their models and ours, the resource rental cost is
considered in this paper as well, since it is a major part which affects the profit of
service providers.

Proposed System:
Proposed a multi‐tier Cloud negotiation model consisting of: 1) user tier (comprising
consumers and brokers represented by CAs and BAs, respectively), 2) service tier
(comprising service providers represented by service provider agents (SAs)) and 3)
resource tier (comprising resource providers represented by resource agents (RAs)). In
negotiation activities were carried out between CAs and BAs, between BAs and SAs, and
between SAs and RAs. Agents in adopted the time‐dependent strategy with fixed
concession rates, but market‐oriented issues such as outside options and rivalry were not
considered. Empirical results in section 3.4 show that the ACR‐MSC strategy in this work
achieved significantly higher utilities than the time‐dependent strategy without sacrificing
success rates in negotiation. Additionally, mathematical proofs show that agents adopting
the ACR‐MSC strategy negotiate optimally. Moreover, in game‐theoretic issues such as
InterCloud coalition formation, equilibrium strategies, and fair division of payoff were
not considered. The agent‐based testbed in consists of PAs and CAs that act as
intermediaries between Cloud resource providers and consumers, respectively, and a set
of BAs that connects resource requests from consumers to advertisements from providers.
Problem Statement: A profit maximization function is defined to find an optimal
combination of the server size R and the queue capacity K such that the profit is
maximized. However, this strategy has further implications other than just losing the
revenue from some services, because it also implies loss of reputation and therefore loss
of future customers. In , Cao et al. treated a cloud service platform as an M/M/m model,
and the problem of optimal multiserver configuration for profit maximization was
formulated and solved. This work is the most relevant work to ours, but it adopts a single
renting scheme to configure a multiserver system, which cannot adapt to the varying

market demand and leads to low service quality and great resource waste. To overcome
this weakness, another resource management strategy is used in , which is cloud
federation. Using federation, different providers running services that have
complementary resource requirements over time can mutually collaborate to share their
respective resources in order to fulfill each one’s demand . However, providers should
make an intelligent decision about utilization of the federation (either as a contributor or
as a consumer of resources) depending on different conditions that they might face,
which is a complicated problem.
Scope: Contributing the game‐theoretic foundations for analyzing and specifying the
interactions of a society of agents in an InterCloud, this work has only taken the first step
towards designing an intelligent InterCloud. The author hopes that this work will inspire
others to take up future challenges of realizing and implementing the ideas and solution
concepts in the paper.
Architecture:

Implementation of modules:
(1)CONSUMER-TO-CLOUD INTERACTIONS:
Negotiation between each pair of CA and PA is carried out by making proposals in
alternate rounds. Unlike Rubinstein’s alternating offers protocol [16], which is a bilateral
(one‐to‐one) negotiation protocol, in the multilateral consumer‐to‐Cloud interaction
protocol, multiple CA‐PA pairs can negotiate deals simultaneously. Each CA
(respectively, PA) can negotiate with multiple PAs (respectively, CAs) at the same time.
Making proposals: When an agent makes a proposal, it proposes a deal from their space
of possible deals. These consist of the most desirable price, the least desirable (reserve)
price, and prices in between. An agent proposes its most preferred deal initially. If no
agreement is reached, negotiation proceeds to the next round.
(2) CLOUD-TO-CLOUD INTERACTIONS:

InterCloud is a federation of Clouds, economic encounters among Clouds can be
modeled as a coalition game. In an InterCloud coalition game (definition 4.1), the
players are the Clouds (represented by hCAs and fCAs) that cooperate with one
another by drawing upon each other’s resources to satisfy consumers’ demands,
collectively generate more profit for the coalition, and share their total profit. Two
major issues in InterCloud interaction are: 1) How does each Cloud choose its
coalition partners? and 2) How should the coalition divide its payoff among the
players? Within an InterCloud, each self‐interested Cloud negotiates and
establishes agreements with other Clouds to meet its own objectives and to
optimize its own payoff. The agent‐based Cloud‐to‐Cloud interaction protocol
specified in algorithm 4.1 consists of four stages: 1) announcement of the
availability of resource capacities, 2) bidding for the priority right to acquire
resource capacities of other Clouds, 3) making offers for sharing the payoff
generated by the InterCloud coalition, and 4) acceptance or rejection of offers.
(3)Adaptive Concession Rate:
Consumers in a Cloud market compete for computing services and Cloud providers
compete to provide services, a market‐oriented approach taking into account the demand
for and supply of Cloud services is appropriate. Bargaining with deadlines: Since
consumers generally have deadlines in acquiring computing resources to execute jobs and
Clouds also have deadlines for scheduling their resources and executing jobs, both CAs
and PAs are programmed to make concessions with respect to time. Both CAs and PAs
are designed with three classes of time‐dependent concession making strategies: i)
conservative (conceding slowly by maintaining the initial price until an agent’s deadline
is almost reached), ii) conciliatory (conceding rapidly to the reserve price), and iii) linear
(conceding linearly).
(4)Minimally Sufficient Concessions:

Even though making a large amount of concession may increase the probability of
reaching an agreement, doing so is inefficient because an agent “wastes” some of
its utility. Nevertheless, if an agent makes too small an amount of concession, it
runs the risk of not reaching any agreement with its opponent eventually. Hence,
making minimally sufficient concession [19] is a desirable property of negotiation
agents. Using the ACR‐MSC strategy, an agent negotiates optimally by making an
amount of concession that is minimally sufficient. The general idea is that for a
given market situation, an agent adopting the ACR‐MSC strategy strives to attain
the highest possible utility while maintaining a minimum probability of reaching
an agreement.
(5)Bargaining theory:
Bargaining or haggling is a type of negotiation in which the buyer and seller of a
good or service debate the price and exact nature of a transaction. If the bargaining
produces agreement on terms, the transaction takes place. Bargaining is an
alternative pricing strategy to fixed prices. Optimally, if it costs the retailer nothing
to engage and allow bargaining, he/she can divine the buyer's willingness to spend.
It allows for capturing more consumer surplus as it allows price discrimination, a
process whereby a seller can charge a higher price to one buyer who is more eager
(by being richer or more desperate). Haggling has largely disappeared in parts of
the world where the cost to haggle exceeds the gain to retailers for most common
retail items. However, for expensive goods sold to uninformed buyers such as
automobiles, bargaining can remain commonplace.
(6)Consumer Module:
In this project consumer module should perform the following tasks
Cloud consumer:
 .consumer registeration.
 consumer login.
 consumer send request to consumer agent for accessing cloud.
 view status of the request.
 after accepting the request consumer upload his files in to cloud.

(7)Cloud Agent Module: In this project consumer module should perform the
following tasks
Consumer agent:
 .consumer registeration.
 consumer agent login.
 consumeragent view the consumer requests.
 forward consumer requests to corresponding cloud brokers.
 view all the requests and responses of cloud brokers.
Conclusion
The significance and novelty of this work are that it is the earliest work to propose
an agent‐based InterCloud economic model for analyzing two types of interactions
in an intelligent InterCloud: 1) consumer‐to‐Cloud interactions and 2)
Cloud‐to‐Cloud interactions. Being the first to devise 1) best response strategies
for InterCloud coalition formation that converge to both a subgame perfect
equilibrium and the Shapley value payoff and 2) an optimal multilateral
consumer‐to‐Cloud negotiation strategy, this work provides game‐theoretic
solutions that lay the essential mathematical foundations for InterCloud economics.
On this account, this work advances the state of the art in many ways as follows.
From the perspective of Cloud computing, this work contributes a new branch of
knowledge for realizing the InterCloud vision. Being the first of its kind to provide
both 1) bargaining game and coalition game solution concepts in an InterCloud and
2) agent‐based interaction protocols for automating consumer‐to‐Cloud and
Cloud‐to‐Cloud interactions, this work is an important milestone in introducing
agent‐based InterCloud economics and agent‐based InterCloud interactions as new
frontiers in InterCloud research.

Agent based interactions and economic

More Related Content

Viewers also liked (11)

Similar to Agent based interactions and economic (20)

Recently uploaded (20)

Agent based interactions and economic