Swarm Based Cross Layer Optimization Protocol for WMSN

ABSTRACT


RELATED WORK
In wireless sensor networks most routing protocols consider energy saving as the main objective. But due to addition of audio, video and imaging sensors additional challenges are exposed. Luis Cobo, Quintero and Samuel have proposed ant based routing protocol for WMSN. Here they have used ant based multi QoS routing metric. It is stated that this algorithm has better convergence and provides better QoS for multiple types of QoS. [1] Abazeed and Saleemhave proposed that improved ACO to search for best path that are satisfied with multimedia traffic requirement. To make best decision the weightage is given to energy consumption and queuing delay. At MAC layer dynamic duty cycle assignment is proposed to satisfy its goals. Result shows that it has achieved better delivery ratio and energy consumption has been reduced. [2] In CRWMSN proposed by Zara Hamid, Faisal Bashir and Jae Young Pyun many challenges related to delay and errors are presented. The protocol uses a routing metric based packet service time, channel utilization, hop count and energy of a node. Results shows that end to end packet latency has been decreased and throughput of the system is increased. [5] MatteoCesana, NestroTiglao, Jose M., PetiaTodorova has presented the main characteristics and requirements of real time multimedia monitoring applications and then highlighted key research directions. It is suggested that application centric cross layer optimization is essential to achieve the required QoS in most energy efficient way. [6] Islam Almalkawi, ManelZapaa and Jamal Al-Karaki has exploited the hierarchical structure of powerful cluster heads and the optimized multiple paths along with the adaptive scheduling to support reliable, high throughput and energy efficient multimedia transmission in WMSN.
Correlation characteristics and functionalities between two layers have been exploited to maximize the network performance with minimum energy consumption. [7] An Ant based protocol designed for WMSN is ASAT (An ant based service aware routing algorithm for WMSN is presented by Y. Sun, Lui, Tung. This protocol defines three different parameters of services as event driven, data query and stream query services. ASAR achieves better performance in terms of better convergence and provides better QoS for multiple types of services. [8] 3.
CROSS LAYERED AntSenseNet PROTOCOL Ant based cross layer routing protocol is implemented to optimize the QoS parameters. AntSenseNet protocol is hierarchical protocol based on ant colony optimization protocol. Ant Sense Net algorithm is divided into three steps: In first step it clusters the nodes in colonies. Then it finds the route between clusters that meet the requirements and lastly it forwards network traffic over the previously discovered route.
WMSN Model: Network is created with multimedia sensor nodes with three types of sensors including audio, video and temperature sensor nodes.
Network simulator 2 is used to develop and simulate the performance of the Ant based Cross layer protocol. Throughput, lifetime, jitter and packet delivery ratio are used as the performance measures to evaluate the performance. The following assumptions are made for this sensor network: Once network is created with multimedia sensors nodes, then the network is partitioned into clusters.Clustering process is fully distributed among all sensor nodes. Clustering algorithm aims to achieve saving of network resources by selecting the nodes which are rich in resources also ensures network connectivity resulting in maximizing network lifetime.
In order to form the cluster, each node needs to calculate the probability and pheromone value of its own. Available memory of a node and remaining energy of node is used to compute the pheromone value of the node. double re = node_->energy_model()->energy() / node_->energy_model()->initialenergy(); pheromone_ = pow(ma(),a) * pow(re,b); Table 1. Network Parameters to simulate Routing Mechanism Once pheromone value is computed then it is broadcasted among all ofitsneighbour with the hello message. Neighbour nodes receive the pheromone value and stores in its neighbor table and the probability of the node is estimated as the ratio between own pheromone/total pheromone value of its neighbor. After calculating probability value of the node, each node initiated to form cluster by invoking clustering function. It will check the probability value of neighbor with own probability, if own probability is maximum among all the neighbours then that node become cluster head (CH).
CH node send cluster message to all of itsneighbour node with energy, pheromone and probability value. Receiver node, updates the status of the head as CH in the neighbor table.If it is already member of the another cluster, then it checks the maximum probability of heads and initiates the join timer to send join message to the maximum probability cluster head node and sends leave message to old cluster head.If the current node is cluster head and probability value of own is greater than neighbour node then CH node send resign message to its one hop neighbour nodes.
Once cluster formation completes, ant procedure is invoked in all sensor.AntSenseNet protocol is based on Ant colony optimization protocol used to discover and maintain the path between source and sink. There are three phases of AntSenseNet protocol: Forward ant phase, Backward ant phase and route maintenance phase. Once the cluster process ends route discovery process starts. Each sensor waits for announcement from the base station. Once announcement is received from the base station then it checks for route to reach base station in the routing FANT is rebroadcasted until message reaches the base station. When the forward ant reaches the base station evaluation of the path is carried out. Once it reaches the base station then base station computes path parameter such as delay energy, packet loss. These parameters are compared with standard parameters set for the particular application. If these values did not match with the standard values then that path is discarded. If FANT values fulfill the requirement of particular application, base station generates backward ants and sends these BANT message in the reverse direction. BS calculates the pheromone value of the route and unicast with the BANT in the reverse route. Intermediatenodes receive and unicast the message to source node and also it updates the pheromone value of the ant nodes in the path.Intermediate nodes create the route in the routing table to reach the base station via ant nodes.Base station maintains Ant release timer to release the ants in the network.
Whenever this timer expires, BS calculates the number of ants to be released in the network.Each node maintains maintenance timer, and checks for traffic load, if it reaches the threshold then it send MANT message with pheromone value.
Receiver node updates the pheromone value with the new pheromone value. From the interfacequeue, mac layer and linklayer, packet delay in terms of queue_delay, transmission_delay, network_delay and mac_delay are estimated, From these values, packet service time is estimated. T_PST = T_net + T_queue + T_mac + T_trans; From MAC idle time and last transmission time, T_PST value is updated. T_Pst = Tr -Tl During data transmission, based on the packet service class, delay for each hop is calculated as d_hop = ch->PST / ch->s_PST * ch->pkt_class; di = (CURRENT_TIME -ch->enq_time); eta_i = di / d_hop; In order to maintain the route, FANTS are generated periodically in order to update the path periodically. During route maintenance congestion and lost link problem are also taken into consideration in order to minimize the packet loss and to improve the throughput.
In this protocol cluster head forwards the data following the maximum pheromone value. In this protocol sensor data is in various forms such as audio data, video data, scalar data. Hence packet scheduling policy is considered for different types of data. Here each CH classifier checks the type of data packet and then it is sent to the appropriate queue and scheduler organizes this data according to type and priority.

RESULTS& ANALYSIS
Here, Different QoS parameters are analyzed for AntSenseNet protocol and Cross layer based AntSenseNet protocol. There are two types of nodes: scalar sensor nodes and multimedia sensor nodes. Radio range of the nodes is 100m. Nodes are randomly distributed in the network having size of 500m X 500m. Modified version of 802.11 MAC protocol is used which uses priority queue and handles multimedia traffic very efficiently. Cross layering between routing, MAC and physical layer is done in order to consider multi priority traffic. 100 nodes are randomly distributed in a square area. Simulation is carried out for 50 Seconds. Initial energy of the nodes was 100 Joules. Simulation is carried out after clustering. Different performance metrics are taken into consideration: Packet delivery ratio, Throughput and Number of packets dropped. Routing packets includes control packets which are used for rout discovery, route maintenance and pheromone updation.
Network is moderately loaded. Two traffic classes are produced: multimedia traffic with a size of 1024 byte packets and scalar traffic having size of 32 byte packets. Multimedia traffic has higher priority than scalar traffic. Fig 1 shows simulation time vs packet delivery ratio of AntSenseNet protocol and Ant based cross layer protocol. It is constant for different simulation time as well as for different coverage ranges. Packet delivery ratio is better for ant based cross layer protocol than the AntSenseNet protocol.
This result is obtained due to two level scheduling that has been used. This two level scheduling avoids collisions and minimizes the interference also. Also due to the multipath transmission this ratio is improved. Due to the clustering process link failure is minimized and coverage has been increased.   Our proposed protocol achieves higher throughput as shown in the figure 4, 5and 6. This is due to the multipath scheduling algorithm that has been used in the protocol. Multiple paths were discovered when path to destination breaks. Packets could reach to destination without new path discovery process. Due to this packet delivery ratio increases and number of packets dropped decreases.
Ant based crosslayer protocol has less number of packets dropped due to multipath routing protocol. Due to clustered network architecture it has less communication overhead as most of the nodes in the network have to communicate with their cluster heads and cluster head communicates with other group heads. Due to this energy required is less which results in increase in lifetime. Alos data aggregation takes place at cluster heads which results in decrease in energy consumption

CONCLUSION
WMSN has issues such as coverage, connectivity, network lifetime and scheduling & data aggregation. In this paper ant based cross layer protocol is presented. Cross layering between routing and MAC layer is done in order to exploit the correlation and functionality between different layers. Also multipath routing protocol has been implemented along with scheduling and queuing model to improve network lifetime and overall throughput of the network. Our proposed protocol uses scheduling protocol which maintains the minimum end to end delay suitable for different data. It achieves high throughput and packet delivery ratio by avoiding collisions and interference. The simulation results shows that proposed protocol achieves better performance than AntSenseNet protocol.
In future we will focus on the clustering of the nodes in order to optimize end to end delay. Also we will concentrate on the on the routing model we have used in proposed protocol in order to improve the energy efficiency and network lifetime.