بکار گیری هوش مصنوعی در بهبود شبکه های حسگر بی سیم قسمت دوم
3. Artiﬁcial Intelligence and Multi-Agent Systems
lassical rtiﬁcial ntelligence aimed at emulating within computers the intellectual and interaction abilities of a human being. he modern approach to rtiﬁcial ntelligence ( ) is centered around the concept of a rational agent. n agent is anything that can perceive its environment through sensors and act upon that environment through actuators (ussell orving, ).n agent that always tries to optimie an appropriate performance measure is called a rational agent. Such a deﬁnition of a rational agent is fairly general and can include human agents (having eyes as sensors, hands as actuators), robotic agents (having cameras as sensors, wheels as actuators), or software agents (having a graphical user interface as sensor and as actuator). rom this perspective, can be regarded as the study of the principles and design of artiﬁcial rational agents.
However, agents are seldom stand-alone systems. n many situations they coeist and interact with other agents in several different ways. amples include intelligent eb software agents, soccer playing robots, e-commerce negotiating agents, computer vision dedicated agents, and many more. Such a system that consists of a group of agents that can potentially interact with each other is called a Multi-gent Systems (MS), and the corresponding subﬁeld of that deals with principles and design of multi-agent systems is called istributed ( ).
4. Wireless Sensor Networks and Artiﬁcial Intelligence
n intelligent sensor is one that modiﬁes its internal behavior to optimie its ability to collect data from the physical world and communicates it in a responsive manner, to a base station or to a host system. he functionality of intelligent sensor includes: self-calibration, self-validation, and compensation. he self-calibration means that the sensor can monitor the measuring condition to decide whether a new calibration is needed or not. Self-validation applies mathematical modeling error propagation and error isolation or nowledge-based techniues. he self-compensation maes use of compensation methods to achieve a high accuracy. he types of artiﬁcial intelligence techniues widely used in industries are: rtiﬁcial eural etwor (), uy ogic and euro-uy. ntelligent sensor structures embedded in ireless Sensor etwors result in wireless intelligent sensors. he use of rtiﬁcial intelligence techniues plays a ey role in building intelligent sensor structures. Main research issues of the Ss are focused on the coverage, connectivity networ lifetime, and data ﬁ-delity. n the recent years, there has been an increasing interest in the area of the rtiﬁcial ntelligence and istributed rtiﬁcial ntelligence and their methods for solving Ss constrains, create new algorithms and new applications for Ss. esource management is an essential ingredient of a middleware solution for S. esource management includes initial sensor-selection and tas allocation as well as runtime adaptation of allocated tasresources. he parameters to be optimied include energy, bandwidth, and networ lifetime. n this par
ticular case Distributed Independent Reinforcement Learning proposed the use of collective intelligence in resource management within WSNs (Shah et al., 2. inall, intelligent net wor ing and collaborative sstems are also proposed as components for WSNs’ enhancement.
5. Multi-Agent Based Simulation
S refers to the simulation aim at modeling the behavior of agents in order to anale their interactions and conse uences of their decision ma ing process. ence, a global result is closel determined b agents’ interactions. In practice, S models are used to repre sent and understand social sstems (onte et al., , moreover to evaluate new strategies of improvement and politics on different ind of sstems. Due to S is a recentl area, there are actuall few techni ues and tools for its development. In fact, some contributions come from sstem simulation, software engineering and agentoriented software engineering (S. acing this constrain, a methodolog was proposed b IDI research group from National niversit of olombia, which deﬁnes several stages and artifacts for ever phase of a software lifeccle (oreno et al., 2. his methodolog allows the representation of main characteristics of the distributed sstem, including e aspects such as organiation, reasoning, communication, and coordination mechanism, among others. he main function of WSN simulators is to emulate a WSN operation and simulate entire characteristics of hard ware for each node in simulated WSN, instead of providing strategies to do a deploment. he fundamental idea is to propose a model that enables a highl distributed sensor networ to behave intelligentl as a multiagent sstem. It is important to note that most simulators are used to simulate a speciﬁc sstem, be a S or a WSN, but not both of them. esides, it is needed to identif the relationships eisting between agents and sensor nodes for getting intelligence from the multiagent sstem and monitoring from the WSN. rom WSNs’ point of view, S provides understanding on WSNŠs performance and networ autonomous capabilities when acting as an agents societ. In this case, agents collaborate together to save and improve resources within the WSN. inall, S can highl contribute to deﬁne de ploment strategies and operation politics related to the simulated application.
6. Multi-agent Model proposal
odel proposal is a ultigent hbrid model to simulate the deploment of software agents over an WSN, this is done b a laered architecture that utilies deterministic models of hardware with agent based intelligence, in order to evaluate different strategies, such as dif ferent agents for a speciﬁc application. It utilies mobile agents to control networ resources and facilitate intelligence. In order to get this, it is used principal deterministic models for WSN performing, such as, protocol model, which comprises all the communication protocols and their operation usuall depends on the state of the phsical platform of nodes, phsical model, which represents the underling hardware and measurement devices, media model, which lin s the node to the real world through a radio channel and one or more phsical channels, batter model that is responsible for chec ing if the node has ehausted its batter through computing power consumption of the different components, among others (gea Lope et al., 2. oreover, it is added the topolog and phsical variables according to the application that is going to be simulated. hen, it is used software agents to perform all tas s re uired b the application stud case.
6.1 Simulation Models for WSN
Present simulation models try to represent how a WSN works. For example, Egea-Lopez at al., in Egea-Lopez et al. ( hae proposed a general simulation model taking into aount urrent omponents o a WSN simulator. ene, there are seeral deterministi models to represent hardware, enironment, power, radio hannels, among others. hese models are use ul in the way o knowing aout how a WSN per orms in a real li e ut they do not o er the potential o ealuating di erent strategies o deployment, moreoer, the simulation nodes numer is really ar o a real network, due to salaility is a eted y all reuired proessing to simulate omplete hardware.
Later, a new propose is presented y heong in heong ( . Some strengths o this work are the use o di erent simulation tools whose are already deﬁned or WSN Leis et al. (, and it permits a direted implementation rom simulation. oweer, heong proposes a programming paradigm ased on ators, whose are a onept etween oets and agents. tors are oets with data ﬂow or ommuniation, ut they are not aware o its enironment neither ale to take deisions or ating.
nother approah is presented y Wang and iang in Wang et al. (, where is presented a strategy to ontrol and optimize resoures in a WSN through moile agents. ptimization o resoures suh as, power, proessing and memory o deies is done, ut it is not deﬁned how deies and agents are related or getting this optimization.
6.2 Model Proposal
t is proposed a ulti-gent hyrid model to simulate the deployment o so tware agents oer any WSN, this is done y a layered arhiteture that uses deterministi models o hardware with agent ased intelligene, in order to ealuate di erent strategies, suh as di erent agents or a speiﬁ appliation.
We aim to utilize moile agents to ontrol network resoures and ailitate intelligene. n order to get this, it is used the prinipal deterministi models speiﬁed y Egea-Lopez et al. (, these models set eatures, suh as, plat orm o nodes, power onsumption, radio hannel and media. oreoer, it is added the topology and physial ariales aording to the appliation that is going to e simulated. Finally, it is used so tware agents to per orm all tasks reuired y the appliation study ase. elow is presented three di erent layers that let to per orm intelligene through agents oer a WSN.
6.2.1 Hardware Layer
he hardware layer is responsile to spei y all omponents that are related to harateristis proided y hardware and the enironment where network is going to e deployed. ost models o this layer are already deﬁned y the present WSN simulators. elow it is introdued some models that spei y these omponents.
•Node odel his model has een speiﬁed e ore y Egea-Lopez et al. (, where a node is diided y protools, hardware and media. Protools operation depends on hardware speiﬁations and omprises all ommuniations protools o a node. ardware represents the underlying plat orm and measurement deies. nd media, links the node to the ¸Sreal world ˇ