Natural Computing in Mobile Network Optimization

Natural Computing in Mobile Network Optimization

Anwesha Mukherjee (West Bengal University of Technology, India), Priti Deb (West Bengal University of Technology, India) and Debashis De (West Bengal University of Technology, India)
DOI: 10.4018/978-1-5225-0058-2.ch017
OnDemand PDF Download:


Nature inspired computing has been widely used to solve various research challenges of mobile network. Mobile network refers to mobile network, sensor network and ad hoc network. This chapter has focused on the application of nature inspired computing in mobile network. In this chapter, the bio-inspired techniques for wireless sensor network, mobile ad hoc network and mobile cloud computing are discussed. Ant colony optimization is used in sensor network and mobile cloud computing for efficient routing and scheduling respectively. Bee swarm intelligence is used to develop routing schemes for mobile ad hoc network. Bird flocking behavior is used for congestion control in wireless sensor network. The research challenges of bio-inspired mobile network are also illustrated.
Chapter Preview


Nature inspired computing (NIC) has grown up since more than a decade. NIC focuses at the development of new computing techniques with the ideas of how nature acts in different scenarios for solving complicated problems. NIC has become an interesting research area because of the technological growth in computer science. NIC techniques are applied in the field of physics, biology, economy, management, engineering etc. The models may be ant colonies or swarms, they are appropriate to model an intricate and dynamic system (Hu, Eberhart, & Shi, 2003). NIC techniques are flexible and applicable for various problems (Dorigo & Stutzle, 2010). These methodologies are adaptable because they can handle invisible data and are able to learn. The methods can manage the partially complete data set. Thus these methods are called robust. The vast majority of these algorithms are imitative from biological systems in nature.

Particle Swarm Optimization (PSO) is a popular NIC method that is developed by close scrutiny of the swarm behavior of fish and bird (Hu et al., 2003). The theory of particle swarm is basically originated from the simulation of the behavior of the real time living swarm. Observation of choreography of a bird flock leads to the preliminary theory for optimizing problem based on swarm intelligence incorporating nearest neighbor velocity matching, environmental ancillary variables, multi-dimensional search and acceleration by distance (Kennedy & Eberhart, 1995). Like genetic algorithm, it is also an evolutionary technique. In this technique, every possible solution with respect to the problem is considered as particle; hence few particles are chosen randomly to form a swarm. This swarm acts as the real time swarm such as bird flocks, ant colony etc. The behavioral activities of a swarm are reflected through mathematical model that has been applied on the computational swarm seeking for optimal solution for the problem. This is known as Particle swarm optimization technique. Like Genetic Algorithm (GA), PSO is popularly used as a population based optimization tool although PSO has no evolution operators like mutation or crossover (Bonabeau, Corne, & Poli, 2010). PSO is a stochastic optimization approach that is based on population and motivated by communal conduct of fish schooling or bird flocking (Pant, Radha, & Singh, 2007). Bee colony algorithm is discussed based on the honey collecting behavior of bee swarms (Yan, Zhu, Chen, & Zhang, 2013). An enhanced version of PSO is proposed by developing hierarchical interaction topologies and this is referred as Multi-swarm Particle Swarm Optimizer (Chen, Zhu, & Hu, 2010). Using swarm intelligence, Bat Algorithm (BA) is formulated on the basis of the echolocation of bats (Yang, 2010). Echolocation is basically a kind of sonar which is used by bats to perceive prey, evade hindrances and trace their roosting fissures in dark (Yang & Gandomi, 2012). In order to resolve the problem related to multi-objective optimization, BA is modified to form Multi-objective Bat Algorithm (MOBA) (Yang, 2011). Multi-objective optimization problems basically involve multiple criteria in decision making. Such types of problems find application in economics, finance, optimal control, optimal design and radio resource management. MOBA is tested using a subset of relevant test functions to get an efficient solution.

Ant Colony Optimization (ACO) is proposed on the basis of the behavior of ants seeking for paths between the source of food and their colony (Dorigo & Gambardella, 1997). In general, ants wander randomly in search of food and after finding food they revisit the colonies. On the path they release pheromone trials (Colorni, Dorigo, & Maniezzo, 1991). Other ants finding such a path follow the pheromone trials rather than moving randomly. However the pheromone trials get evaporated with time, thus reducing the effectiveness of the search. More the time taken by the ant to traverse the path more is the chance for the pheromone to get evaporated. Therefore when an ant gets a good short path for food in the colony, other ants go behind that path. Positive feedback finally leads all the ants to go after the path. Based on the honey collection behavior, bee swarm algorithm is discussed in (Karaboga, 2005). A survey on the existing NIC techniques is carried out in (Karaboga, Dervis, & Akay, 2009).

Key Terms in this Chapter

Handoff Management: Handoff management is an important area in mobile computing. Handoff management is generally categorized into two types: hard handoff and soft handoff. The resources allocated by the previous cell are first released in case of hard handoff. After that the resources of new cell are allocated to the subscriber in this case. But in soft handoff first the resources of the new cell are allocated to the subscriber and then the resources of the previous cell allocated are freed.

Wireless Sensor Network: Wireless Sensor Network (WSN) is composed of sensor nodes with various sensing abilities e.g. temperature, light, acceleration etc. Sensor nodes collect the status of an object and send to the sink node. These sensor nodes can interact with each other. Wireless sensor networks are used to monitor the status of real time environment. The sensor network consists of a number of sensor devices which are deployed over a geographical area. Sensor network protocols and algorithms possess self-organizing capabilities. WSN nodes encompass the following components: embedded processor, low-power sensing device, power module and wireless communication subsystem. The embedded processor is used for collecting and processing the signal data taken from the sensors. Their processing capabilities are utilized to perform computations locally and convey the required data only which is partially processed. Wireless communication subsystem is used for data transmission. The power source consists of a battery with a limited energy budget. Sensors make a measurable response if any change occurs in the physical condition like temperature, humidity etc. The communication technologies used in wireless communication are RFID, Zigbee, IEEE 802.15.4 etc. Sensor networks consist of different sensor nodes such as acoustic, thermal, radar, seismic and visual able to monitor a variety of condition including humidity, temperature, vehicular movement, soil makeup, pressure, the presence of certain objects, noise levels etc. Sensor nodes can be used for continuous sensing, event detection, event ID, location sensing and local control of actuators. The applications of wireless sensor network can be categorized into military, environment, health, home, disaster relief. Different deployment technologies are used for wireless sensor network: Regular deployment where sensor nodes can be allotted according to a well planned, fixed manner. In such case data is travelled through a predefined path. This deployment is generally used in home networks, industrial sector etc. Random deployment means distributed sensor nodes over a finite area. When the sensor node deployment scheme is not predefined, optimal positioning of cluster head becomes vital for energy efficient network operation. Sensor nodes can be featured with mobility where nodes can move to manage the deployment difficulties and passively moved about by an external force i.e. water, wind and vehicle. This type of sensors is generally used in battle field surveillances, emergency situations i.e. fire, Volcano, Tsunami.

Mobile Ad Hoc Network: A self configured network, infrastructure less mobile network. Here two or more devices are associated wirelessly. If two nodes are situated in the radio range of each other, they can communicate directly. Else they communicate with the help of other nodes. When two nodes communicate in a network, a unique identity is compulsory. The configuration of Mobile ad hoc network (MANET) takes place dynamically in an automated way. Thus it is called as a self-organized network. MANET is an infrastructure less network also where the nodes connect with other wirelessly. In Latin ad hoc means “for this purpose”. In a MANET the nodes move about independent of other nodes hence, the links between nodes are frequently broken and established. The nodes in a MANET forward traffic that is not linked to it. Hence it serves as router. If the nodes are within the transmission range of each other then they can communicate directly, but if they are not then they communicate via other nodes. In a MANET each participating nodes must maintain the necessary information required to route packets. As the nodes are mobile in a MANET, the network topology is changed dynamically. MANETs are also susceptible to various security threats. The features of MANET are: seamless communication between nodes and suitable environment for ubiquitous mobile computing, infrastructure less and self-configured, supports mobility, seamless connectivity is maintained even when the nodes move about from their positions, the nodes act as router and forward unrelated packets, ability to discover nearby nodes, computations take place in a decentralized way, resource availability is limited, wireless connectivity range is limited, and vulnerable to various security threats.

Mobile Cloud Computing: Mobile cloud computing is an integration of mobile computing and cloud computing. Due to the advancement in mobile network, mobile web users are increasing explosively. But the mobile device suffers from poor battery life, limited resource and storage capacity and limited computation power. To overcome these problems, cloud is incorporated into the mobile network services. In mobile cloud computing, the data storage and computation takes place as a whole or partially inside the cloud. As a result, the processing power of mobile device is saved and consequently the battery life is increased. Mobile cloud computing provides mobile cloud applications to a wider broad range of mobile users. Mobile cloud computing is a combinational infrastructure of two domains: cloud computing and mobile computing. Mobile cloud computing is composed of three components: mobile network: it contains the mobile devices such as smart phones, laptops, tablets, PDA etc and the network operators. These mobile devices are connected to the network operator through the cellular base station, satellites or access points; Internet service: it acts as a bridge between the mobile network and the cloud. The requests of users are forwarded via wired or wireless connection to the cloud; and Cloud service: when the cloud controller receives user's request, it processes the required application and returns the result to the user. In cloud, the data center offers the hardware facilities and infrastructure. The services related to storages, servers, hardware, networking components are provided to the user in the Infrastructure as a Service layer. Platform as a Service layer offers various platforms such as .NET, PHP, Java etc. Software as a Service layer provides various software solutions.

Offloading: The remote execution of an application inside the cloud is known as offloading. Offloading can be full and partial. If an application is executed fully inside the cloud at remote server side, this is called full offloading. Else if an application is executed partially inside the cloud, this is known as partial offloading. In this case, rest of the part is executed inside the mobile device. Offloading refers to the data transfer from a digital device to another digital device. It is a solution where computations are migrated to the resourceful computers in order to increase the capabilities of mobile devices. This method is different from the conventional client-server architecture. Client can be of two types: thin client and fat client. In case of thin client, data processing and transformation are performed at cloudlet; the client i.e. the mobile device is responsible only for retrieving and returning the data when it is being asked without any considerable processing. Applications which contains huge algorithm are unable to get processed only by the mobile device. Such applications are required to be executed inside the cloudlet for fast processing. Face recognition and voice to text translation are examples of such application. In case of fat client, the server is responsible for managing the data access only where most of the processing and transformation take place inside the client i.e. the mobile device. The mobile devices which are able to process resource rich applications are called fat clients. Instagram is a popular application which can be installed inside a mobile device. Using this application user can take a photo or video, and then can transform it and post on social networks. In such a scenario small processing takes place inside the cloudlet. This is an example where mobile devices serve as fat clients. For mobile devices cyber foraging is proposed. It is described as a mechanism to augment the computational and storage capabilities of mobile devices through task distribution.

Macrocell: Microcell is the second largest cell in mobile network. It has a coverage area of 500m-2 km. It has a coverage area of 1-20 km. Due to medium coverage area, the power transmitted by a microcell base station is high but less than that of a macrocell base station.

Macrocell: In a mobile network the coverage area of a base station is known as cell. Macrocell is the largest cell in mobile network. It has a coverage area of 1-20 km. Due to large coverage area, the power transmitted by a macrocell base station is very high but at the boundary region the received signal strength is low.

Femtocell: Femtocell is a secure, low power and low cost cellular base station. It is usually allocated in an indoor region. It can serve limited number of users. It is connected to the core network through a security gateway which provides data transmission securely. Femtocell has a very small coverage area of 10-20m. Femtocell is also known as Home Node Base Station as it is generally deployed at indoor region. As the coverage area of a femtocell is very small, the power transmission by a femtocell is very low but the signal strength under its coverage is high.

Heterogeneous Mobile Network: If different categories of cells are used to cover a region, then this is called heterogeneous mobile network. In heterogeneous mobile network two-tier architecture is popular where femtocells are allocated inside the macrocell. Femtocells can be allocated within the microcell also to provide good indoor coverage. In heterogeneous mobile network, picocells are also used along with macrocells, microcells and femtocells. A picocell has a coverage area of 4-200m.

Complete Chapter List

Search this Book: