Data Aggregation Architectures of Wireless Sensor Network !

In a wireless sensor network, sensor nodes are energy constrained, so if all the sensors nodes transmit their sensed data directly to the base station then it consumes a lot of energy of sensor nodes and decreases the network lifetime. Different architectures have been evolved for transmitting the data from source node to destination node. Generally there are two types of architecture for data transmission and data collection in wireless sensor network.
Keywords:data aggregation techniques in wireless sensor network,tree based architechitecture for data transmission in wsn. chain based architecture for dta transmission in wsn, Flat networks for wsn, Hierarchical architecture for wireless sensor network, Spin partocol for data aggaregation in wireless sensor network, grid based architecture for data aggregation in wsn

1. Flat Network Based Architecture

In a flat network each sensor node play same role and has same type of features. In such networks data collection is performed in a data centric manner where the sink node generally send a query message to the sensor node for example via flooding. The sensors which have data matching to the query send response to the sink node.

Some data collection techniques which are used in flat network are as follows:

I.       Flooding and Gossiping

Researcher W.R. Heinzelman [5] proposed adaptive protocols for information dissemination in wireless sensor networks in which flooding technique is used. In flooding each sensor which received the query message from sink, broadcast that message to its neighbors and this process continues until the message reached at destination or maximum number of hops for the packet is reached. Gossiping is slightly enhanced version of flooding where the receiving node sends the packet to randomly selected neighbor, which pick another random neighbor to forward the packet. The gossiping will not broadcast the packets as like flooding and thus overcomes the problem of implosion as any node will be having single copy of the packet to be sent. But however, this may induce delays in transmitting the data to all other node [6].

II.    Direct Diffusion

Direct Diffusion [7] is a data centric and application aware paradigm in the sense that all the data generated by the sensor nodes is named by the attribute value pairs. In direct diffusion the base station requests data by broadcasting task to be performed by the network. This task is defined using a list of attribute pairs such as name of object, interval duration and geographical area.

III. Sensor Protocols for Information via Negotiation (SPIN)

The idea behind the SPIN is to name the data using high level descriptors or Meta data. Before transmission Meta data is exchanged among the sensors via data advertisement mechanism which is the key feature of SPIN [8]. Each node upon receiving new data advertises it to its neighbors. It is a variation of direct diffusion. In rumor routing the idea is route the queries to the nodes that has observed a particular event rather than flooding the entire network to retrieve information about occurring events.

IV. Gradient Based Routing

In this technique the key idea is to memorize the number of hops when the queries are diffused throughout the network. Each node calculates the parameter height of the node which is the minimum number of node required to reach the base station. The difference between the node height and that’s of its neighbor is called the gradient of the link.

2.        Hierarchical Network Based Architecture

In hierarchical wireless sensor network nodes are arranged at different level for transmitting data to the sink node. Hierarchical network may be in the form of cluster, tree, grid and chain. In energy constrained sensor networks of large size it is inefficient for sensors to transmit the data directly to the sink. In such scenarios sensors can transmit data to a local node at upper level or cluster head which collect the data from all the sensors in its cluster and transmits the data to the sink.  This results in significant energy savings for the energy constrained sensors. Different hierarchical Based architectures [9] are as follows:

I.       Cluster Based Architecture

In cluster based architecture for data transmission and data collection at first different clusters of nodes are formed, after that a cluster head is selected for each cluster and other sensor nodes in the cluster send the sensed data to cluster head. This cluster head transmit the collected data to the sink.

Basic structure for cluster based architecture is shown in Figure 1.  In figure there are three clusters of sensor nodes each cluster has a cluster head which receives the data of all other sensor nodes of the cluster and send collected data to base station. Sensor nodes shown in gray color inside the each cluster are normal sensor nodes and sensor node shown in blue color inside each cluster is cluster head of that cluster. Normal sensor nodes N1, N2, N3 and N4 inside Cluster1 send their data to cluster head of Cluster1. Similarly sensor nodes N5, N6, N7 inside Cluster2 and sensor nodes N8 and N9 inside Cluster3 send their data to cluster head of Cluster2 and cluster head of cluster3 respectively.

Figure 1: Cluster Based Architecture

II. Chain Based Architecture

 In chain based data transmission and data collection architecture each sensor sends data to its closer neighbor. All sensors are structured into a linear chain for data collection. S. Lindsey and C. Raghavendra [10] proposed a Power Efficient Data Gathering Protocol for Sensor Information system (PEGASIS), which is a chain based protocol in which nodes can form a chain using greedy algorithm. Greedy chain formation assumes that all the nodes have the global information about the network. The farthest nodes from the sink initiate the chain formation and each step the closest neighbor of a node is selected as its successor in the chain.

Figure 2: Chain Based Architecture
Figures 2 represents the chain based architecture in which sensor node P and S initiates the chain formation and send their data to their neighbor nodes Q and R respectively and these nodes send data to Leader node. Leader sensor node is the node which is very near to Sink.

III. Tree Based Architecture

In tree based architecture for data transmission and data collection nodes are arranged at different levels and tree is rooted at sink node and source nodes work as leaves node.  A node has a parent node and it forwards data to it. Data flow start from source node to the sink through parent nodes. One of the important aspects of tree based approach is to construct an energy efficient data collection tree. 
     An example to illustrate the data transmission and data collection using tree based architecture in wireless sensor network is shown in Figure 3.  It is shown in figure that source nodes A and B forward their sensed data to their parent node E. Similarly source nodes C and D send their sensed data to their parent node F. Parent node E and F forward the collected data to the sink.

Figure 3: Tree Based Architecture

References :

[5] W.R. Heinzelman, J. Kulik, H. Balakrishnan, Adaptive protocols for information dissemination in wireless sensor networks, Proceedings of the ACM MobiCom’99, Seattle, Washington, PP 174–185, 2003.

[6] Hedetniemi, S. M., Hedetniemi, S. T. and Liestman, A. L. A survey of gossiping and broadcasting in communication networks. Networks, Annual symposium on theritical aspects of computer Science, Vol 18, PP- 319–349, 2001.

[7] C. Intanagonwiwat, R. Govindan, D. Estrin, J. Heidemann, and F. Silva, Directed Diffusion for Wireless Sensor Networking, IEEE/ACM Transactions on Networking, vol. 11, pp. 2–16, Feb 2003.

[8] SPIN - W. Heinzelman, J. Kulik, and H. Balakrishnan, Adaptive protocols for information dissemination in wireless sensor networks, in the Proceedings of the 5th Annual ACM/IEEE International Conference on Mobile Computing and Networking (MobiCom’99), Seattle, WA, August 1999.

[9] W. Heinzelman, A. Chandrakasan, and H. Balakrishnan, An Application-Specific Protocol Architecture for Wireless Microsensor Networks, IEEE Trans. Wireless Comm., vol. 1, no. 4, pp. 660- 670, Oct. 2002.

[10] S. Lindsey and C. Raghavendra, PEGASIS: Power-Efficient Gathering in Sensor      Information Systems, 9th ACM Conf. Computer andCommunication Security, PP. 41-47, 2001.