CONGESTION CONTROL FOR A ULTRA-WIDEBAND DYNAMIC SENSOR NETWORK USING AUTONOMIC BASED LEARNING

The physical conditions of the area of interest is being collected at the central location using a set of dedicated sensors that forms a network is referred to as Wireless Sensor Network. A dynamic environment is required for a secure multi-hop communication between nodes of the heterogeneous Wireless Sensor Network. One such solution is to employ autonomic based learning in a MAC Layer of the UWB TxRx. Over a time period the autonomic based network learns from the previous experience and adapts to the environment significantly. Exploring the Autonomicity would help us in evading the congestion of about 30% in a typical UWB-WSNs. Simulation results showed an improvement of 5% using Local Automate Collision Avoidance Scheme (LACAS-UWB) compared to LACAS. Key words: Autonomic Network Architecture, Dynamic Environment, LACAS, Stocastic Model, Ultra Wide-band, Wireless Sensor Networks. Cite this Article: Sanjay K N, Shaila K, Venugopal K R, Congestion Control for a Ultra-Wideband Dynamic Sensor Network Using Autonomic Based Learning, International Journal of Computer Engineering and Technology 10(5), 2019, pp. 20-37. http://www.iaeme.com/IJCET/issues.asp?JType=IJCET&VType=10&IType=5 Congestion Control for a Ultra-Wideband Dynamic Sensor Network Using Autonomic Based Learning http://www.iaeme.com/IJCET/index.asp 21 editor@iaeme.com 1. INTRODUCTION In collaborative space-timing task requires a low cost integrated sensing, communicating and computing nodes that involves an innovative technology in wireless networking, array processing and microelectronics. In the areas of embedded systems, networking, multi-agent systems and pervasive computing WSN finds a significant consideration due to the real time scenarios like environment monitoring, disaster relief. With the combination of large number of static sensor nodes the distributive sensing would be achieved in WSNs [1]. The unique functioning of WSNs can be characterized and the effective use of the communication protocol itself is mandatory and demands for the cross-layer design. One such approach is to combine the Local Automate and Autonomic Network Architecture at the MAC level of the network [2]. This leads to self-healing network that are energy efficient MAC with self-organizing and fault-tolerant routing protocols that involves distributed algorithms. MAC rules have been developed to minimize interference and packet collisions that includes [4], [5], [6]: optimizing the channel access, packet transmission and retransmission methods; packets lengths; modulation and coding; transmission powers; etc. are few of the well-known algorithms that are used till date [8], [9]. These techniques, are not well-suited to the WSNs due to the addressing issues. Thus, a serious paradigm shift in MAC designs is required. Therefore, the decentralized character of a typical WSNs has to be rolled out that complicates with any number of attempts that are required to attain a network wide synchronization [10], [11]. Then, scheduling a data load in WSNs generally becomes very low: with the generated traffic in a highly directed data-processing sink unit‟s that exhibits a converge communication pattern. This needs special observation in the design process since, the nodes are close to the sink. The sink node has to manage more traffic with the available nodes in the perimeters using local automata [12], [13]. In WSNs MAC design two parameters play a vital role namely, number of nodes and number of two hop neighbor nodes. The processing capabilities is reduced due to low complexity of the nodes. While the average scheduling and end-to-end data reporting times are delayed because of limited buffer size [14]. The energy estimation is the largest design limitation of WSN for long network runtimes. In [15], the nodes are switched-off all the time irrespective of the intrinsic current leakage in the battery that limits the life-time upto 10-15 years depending on the operating temperature. This provides maximum network life-time with battery powered nodes. In specific sensor applications, the utilization of energy is controlled by the node‟s radio consumption and in this particular case the network has to prefer UWB. It can be shown that when the node is in sleeping mode power consumption is negligible that enhances the life time since the radio is controlled by the MAC. Non-rechargeable battery; rechargeable battery with regular recharging (e.g. sunlight); rechargeable battery with irregular recharging (e.g. opportunistic energy scavenging); capacitive/inductive energy provision (e.g. active RFID); etc. [15] are few of the power mechanisms that are considered while increasing the life time of the battery. This has an influence on the choice and design of MAC protocol with local automata [16-18]. A large number of ultra-small autonomous UWB devices are considered in our work wherein the sensor node is equipped with the integrated sensors, data processing capabilities and short-range radio communications. The data communication between the nodes are Sanjay K N, Shaila K, Venugopal K R http://www.iaeme.com/IJCET/index.asp 22 editor@iaeme.com forwarded to the specialized gateway nodes. Two alternative routing approaches have been considered for sensor networks namely, flat multi-hop and clustering [20-22]. The data communication between nodes are achieved via specialized gateway nodes using flat, multi-hop and clustering routing approaches. To minimize the costs the sensor data can be combined and compressed inside the node or cluster of nearby nodes and reduces the payload of the data packets [23]. The challenge is to manage the packet‟s overhead condition that is significant in WSN. The existing approaches mainly focusses on routing and destination identification issues [25]. A critical problem still exists that is the overhead of the MAC header or the MAC address. In current approach the unique identifiers are used which are of same size or larger than packet payload that shows the important source of energy consumption. The free space available in the cellular system contains the address agnostic and the addresses that are present in the data packets. This approach for the MAC addressing in the sensor network results in the energy savings [26-30]. The ANA architecture consisting of two layers of co-ordination namely, lower task execution layer and higher task allocation layer as shown in Figure 1. The dotted line shows the detail Autonomic Network Architecture that differs from existing layered architectures with multi-hop coordination that adopts a reactive method [31]. The following characteristics are exhibited while employing ANA with LACAS systems:  Self-configuring: The sensor netwoks are made adaptive for the dynamic changing environment by the process of task allocation and execution schemes. 
  Self-optimizing: In terms of topology, propagation and interference, the system configurations are autonomously and continually adapted to the traffic profile and network environment.  Self-healing: The system of task distribution is robust to the network failures where task execution is capable of self-repairing unexpected robost formation damage.  Self-protecting: The task performance allows the system to route and negotiate the complicated unforeseen barriers.  Task Allocation for nodes: In the extreme case, the movement of nodes will be restricted due to the less work or no work involved. Thus, the general system performance is adversely affected by a task interference. To minimize the physical interference, in our work the nodes are distributed dynamically.  Complexity of dynamically changing nodes: Existing nodes tend to underuse the sensor inputs that may provide helpful data to coordinate behaviors and choose the most suitable action.  Coalition Formation for Minimalist nodes: Exciting multiagent formation schemes requires complex planning, particular formation schemes that require complex scheduling, explicit negotiation and precise coalition cost evaluation. Thus, they may not be able to operate in actual time in a large-scale sensor network.  Cooperation of Resource-Constrained nodes: Only local, uncertain environmental information with limited communication and sensing capabilities can be obtained from sensor nodes. Congestion Control for a Ultra-Wideband Dynamic Sensor Network Using Autonomic Based Learning http://www.iaeme.com/IJCET/index.asp 23 editor@iaeme.com Figure 1. ANA: Autonomic Network Architecture Motivation The focus of the WSNs design is to provide the assurance of its long existence under specific energy and complexity constraints. The MAC plays an important role in this design because it has control on the active and sleeping state of each node. Therefore, MAC protocols needs the primary design factors like reliability, longevity, fairness, scalability and latency [6]. Congestion is the most important factor to be considered to achieve the good reliability of transmission of data in the network. Due to the congestion in WSN‟s traffic increases, that leads to dissipation of energy in colossal amount in the sensor nodes. This results in loss of packets and it creates unfair and non-reliable flow of packets [11]. In many cases nodes are made to work without any interruption for long durations without replacing the energy sources. Therefore, the key concern of WSNs is to optimize the energy consumption in sensor node [15]. With the help of intermediate nodes and localized control centres all the nodes in a sensor network transmit the data to the sink node [18]. This increases the possibility of congestion at the intermediate nodes. These issues have to be addressed while designing WSNs for any applications [19-20]. Contribution To achieve reliable communication between the nodes at larger rate a Learning based Autonomic Network Architecture is proposed that integrates the UWB sensor network with the complex dynamic environments. The throughput and fault tol