Design a Hybrid System Geno-Neuro-Fuzzy Controller for Dynamic Load Balancing in Wireless Ad- hoc Networks

of 19
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Document Description
Journal of Telecommunications, ISSN 2042-8839, Volume 21, Issue 2, August 2013
Document Share
Document Tags
Document Transcript
  JOURNAL OF TELECOMMUNICATIONS, VOLUME 21, ISSUE 2, AUGUST 2013  12 Design a Hybrid System Geno-Neuro-FuzzyController for Dynamic Load Balancing inWireless Ad- hoc Networks Hussein A.Lafta   Assistant Prof., Babylon University, Iraq Abstract - Congestion at the link and in the nodes is the main cause of a large delay in the ad hoc networks, whereband width is limited. Balancing the work among the network nodes will be one of the best solutions. Once, thesource node may be has selected a set of paths to destination, it can send data to a destination along unloaded pathnodes. Load balanced routing aims to move traffic from the areas that are above the optimal load to less loadedareas, so that the entire network achieves better performance. If the traffic is not distributed evenly, then some areasin a network are under heavy load while some are lightly loaded or idle. Therefore good load balancing algorithmsmust be fast and should not add heavy cost, because complexity of these algorithms in communication channelsincurs ambiguity, causes uncertainty in decision making. A novel approach based on artificial intelligence field suchas neural networks, fuzzy logic, and genetic algorithm is suggested in this work.The integration of these subsystems  gives a system benefits from the advantages of each subsystem and encroaches onthedisadvantages. All the parameters of the controller is tuned and learned by genetic algorithm. This controller is based on back propagationneural network. This can eliminate laborious design steps such as manual tuning, of the membership functions andselection of the fuzzy rules and weights of neural network, which give the neural network a greater ability to generalizeand accelerate the convergence process and prevent the network to get stuck in a local minimum.  Key words : fuzzy system, neural networks, genetic algorithm, loads balancing, ad hoc networks.   1. Introduction Mobile ad hoc networks (MANETs) are formed dynamically by an autonomous systemof mobile nodes that are connected via wireless links without using the existing network infrastructure or centralized administration. The nodes are free to move randomly and organizethemselves arbitrarily; thus, the network’s wireless topology may change rapidly andunpredictably. Such a network may operate in a standalone fashion, or may be connected to thelarger Internet. Mobile ad hoc networks are infrastructure-less networks since they do not requireany fixed infrastructure, such as a base station, for their operation. In general, routes betweennodes in an ad hoc network may include multiple hops, and hence it is appropriate to call suchnetworks as ‘‘multi-hop wireless ad hoc networks’’. Each node will be able to communicatedirectly with any other node that resides within its transmission range. For communicating withnodes that reside beyond this range[9], [11],[13] ,[40],[44],[49].  JOURNAL OF TELECOMMUNICATIONS, VOLUME 21, ISSUE 2, AUGUST 2013  13 Load balancing is a technique for distributing traffic load from the source to the destination getoptimal resource utilization, throughput. Load balancing attempts to maximize systemthroughput by migrating tasks from the overloaded nodes to other lightly loaded nodes toimprove the overall system performance. Loadbalancing is important since node with high loadswill deplete their resources quickly, thereby increasing the probability of network failure.Therefore good load balancing algorithms must be fast and should not add heavy cost, becausecomplexity of these algorithms in communication channels incurs ambiguity, causes uncertaintyin decision making. To solve this problem, artificial intelligence field such as neural networks,fuzzy logic, and genetic algorithm are suggested in this work, because they   don’t use anymathematical model of the system. Neural Network uses input-output relations of the system,Fuzzy Logic uses heuristic knowledge about the   system based on the operator experience, andgenetic algorithm is used for search and optimization [6], [7], [10], [17],[26], [29]. 2 Related Works. Many studies have focused on developing mechanisms for the load balancing problem inmobile wireless ad hoc networks. Load-Balanced Ad Hoc Routing (LBAR) protocol is proposedin [20]. LBAR is very similar to AODV, except that while forwarding the route request (RREQ)messages each node appends its current activity status to the RREQ. The destination uses thisextra information in order to relay the RREP message along the path with least load so that theload is evenly spread across the nodes. In [8], the authors propose load balancing algorithm for wireless access networks. The algorithm maintains a load-balanced backbone tree rooted at theaccess point. The access point is responsible for updating the backbone tree and informationabout load distribution is required at the access point. Also it is assumed in [37] that nodes havemultiple-antennas, so wireless connections between neighbors is modeled as isolated point-to- point links, which makes the topology very similar to the wired-networks. In [12], the authors propose a mechanism of load balancing in ad hoc wireless networks that relies on dissipation of load distribution information throughout the network. If a node is overloaded (serving more loadthan the average load in the network), then it queries it neighborhood for under loaded nodes,and transfers its load to the under loaded node. However the dissipation of load informationincreases the overhead and may cause instability.In [33], the authors study that how alternate path routing, which is a popular load balancingmechanism in wired networks, performs in a wireless ad hoc network. It   is observed that theinterference, caused by channel sharing, significantly reduces the performance of APR inwireless ad hoc networks. An analytical model of load distribution in multi-path routing is presented in [48]. It is shown that the multi-path routing does not provide any benefits in ad hocnetworks, unless the number of paths is large. In [14] the authors show that multipath routingmay achieve significant improvement over single path routing provided the various pathsinvolved in multipath routing are sufficiently disjoint. They use the number of interfering linksexisting between the nodes of two paths as the measure of correlation between the paths andchoose the paths with minimum “interference correlation”. It is shown in [30] that optimum load balancing is a NP-hard problem even for a simple network topology.In Dynamic Load Aware Routing (DLAR) protocol [2],[3], [15],[41] routing load of a route has been considered as the primary route selection metric. The load of a route is defined as thesummation of the load of nodes on the route, and the load of a node is defined as the number of  packets buffered in the queue of the node. To utilize the most up-to-date load information when  JOURNAL OF TELECOMMUNICATIONS, VOLUME 21, ISSUE 2, AUGUST 2013  14 selecting routes and to minimize the overlapped routes, which cause congested bottlenecks,DLAR prohibits intermediate nodes from replying to route request messages.Another network protocol for efficient data transmission in mobile ad hoc networks is LoadAware Routing in Ad hoc (LARA) [46] networks protocol. In LARA, during the route discovery procedure, the destination node selects the route taking into account both the number of hops andtraffic cost of the route. The traffic cost of a route is defined as the sum of the traffic queues of each of the nodes and its neighbors and the hop costs on that particular route. Thus, the delaysuffered by a packet at a node is dependent not only on its own interface queue but also on thedensity of nodes. In routing with load balancing scheme (LBAR) [20], the destination collects asmuch information as possible to choose the optimal route in terms of minimum nodal activity (i.ethe number of active routes passing by the node). By gathering the nodes activity degrees for agiven route the total route activity degree is found. Load Sensitive Routing (LSR) protocol [27]is based on DSR. In LSR the load information depends on two parameters: total path load andthe standard deviation of the total path load. Since destination node do not wait for all possibleroutes, the source node can quickly obtain the route information and it quickly responds to callsfor connections. Correlated Load-Aware Routing (CLAR) [25] protocol is an on-demand routing protocol. In CLAR, traffic load at a node is considered as the primary route selection metric anddepends on the traffic passing through this node as well as the number of sharing nodes.Alternate Path Routing (APR) protocol [32] provides load balancing by distributing trafficamong a set of diverse paths. By using the set of diverse paths, it also provides route failure protection. Reference [6] gives a comparative study of some of the load balanced ad hoc routing protocol.The references [43], [7], [34] select the path that consumes the least energy to transmit a single packet from source to[9] destination, aiming at minimizing the total energy consumption alongthe path. The second one [10],[29] intends to protect the overused nodes against breakdown,aiming at maximizing the whole network lifetime.In [2] the authors presented approach based on fuzzy logic for implementing dynamic load balancing algorithm, their algorithm takes into account uncertainty and inconsistency and shows better response time than round robin and randomize algorithms.The author in [4] was proposed a fuzzy dynamic Load balancing Algorithm for HomogenousDistributed Systems by all nodes in the network. The proposed algorithm utilizes fuzzy logic indealing with inaccurate load information, making load distribution decisions, and maintainingoverall system stability. In terms of control, he proposes a approach that specifies how, when,and by which node the load balancing is implemented. His approach is called centralized butdistributed. 3. Fuzzy logic Fuzzy logic has The ability to deal with uncertainty, ambiguity and vagueness [3], [5],[16], [23],[45]. Fuzzy logic is reliable if the mathematical model of the system to be controlled isunavailable, and for nonlinear, time varying systems. The knowledge is represented linguisticallyin the form of fuzzy production rules (if-then) to emulate human expert. Figure (1) shows thestructure of a fuzzy system. The structure is made of the following components:1-  Input fuzzifier interface consists of fuzzification unit which converts the inputs (crisp value) tothe system into membership degree (fuzzy input) by using suitable membership functions.2- Knowledge base comprises of a data base and fuzzy rule base which characterizes the desiredoutput response applied by means of a set of control rules. Fuzzy rules are linguistic type of IF-THEN statements involving fuzzy sets, fuzzy logic, and fuzzy inference. Linguistic rules  JOURNAL OF TELECOMMUNICATIONS, VOLUME 21, ISSUE 2, AUGUST 2013  15 describing the control system consist of two parts: an antecedent and a consequent block. Theyare usually of the form:IF  X  1   is  A i1   AND  X  2   is  A i2   AND…  X  n   is  A i   n   THEN Y 1 is  B i1 .Where  X  1 … X  2   are inputs, Y  1   is the output and  A in   is the input membership function and  Bi 1   is the output membership function.3-  Inference mechanism (rules propagation)   is the kernel of Fuzzy Logic system which has thecapacity of simulating the human decision making mechanism based on fuzzy concepts andfuzzy control actions.4- Output defuzzifier interface consists of defuzzification unit which converts membership gradesof outputs (fuzzy output) into a crisp value( crisp out put).Figure (1): Fuzzy Logic System ArchitectureDesigning of a fuzzy controller requires a number of trial and error iterations, and eventhen, it is very difficult to ensure that the designed controller is an optimal one. Developing arule base is one of the most time consuming part of designing a fuzzy logic system. 4.Genetic Algorithms Genetic Algorithms (GAs) are search algorithms based on mechanics of natural selectionand natural genetics .Usually it is very difficult to transform human knowledge and experienceinto a rule base of fuzzy logic controller. Genetic Algorithms, have been used tune membershipfunctions (optimal shapes, ranges and number of member functions) and for automatic andonline rule learning because of their robustness and ability to provide global solutions.The standard genetic algorithm proceeds as follows: an initial population of individuals isgenerated at random or heuristically. Every evolutionary step, known as a generation, theindividuals in the current population are decoded and evaluated according to some predefinedquality criterion, referred to as the fitness, or fitness function. To form a new population (thenext generation), individuals are selected according to their fitness. Many selection proceduresare currently in use, one of the simplest being fitness-proportionate selection, where individualsare selected with a probability proportional to their relative fitness. This ensures that theexpected number of times an individual is chosen is approximately proportional to its relative performance in the population.   Thus, high-fitness (‘good’) individuals stand a better chance of ‘reproducing’, while low-fitness ones are more likely to disappear.
Search Related
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!