Cluster Based Hierarchical Wireless Sensor Networks (CHWSN) and Time Synchronization in CHWSN Poonam Yadav IIIT,Allahabad, India yadav poonam@is.iiita.ac.in Nagesh Yadav IIIT,Allahabad, India nagesh@is.iiita.ac.in Shirshu Varma IIIT, Allahabad, India shirshu@iiita.ac.in Abstract— A Cluster Based Hierarchical Wireless Sensor Net- work Architecture is proposed to facilitate more than one application sharing the whole or a part of a wireless sensor network, where each application may have its own network, processing requirements and protocols. Such a network is divided into clusters and the clusters are organized hierarchically. For synchronizing the CHWSN there is the requirement for a cluster based hierarchical time synchronization algorithm. None of the already existing time synchronization algorithms satisfy the needs of time synchronization in our CHWSN architecture. Thus the existing time synchronization algorithms TPSN (Time synchro- nization Protocol for Sensor Networks) and FTSP (Flooding Time Synchronization Protocol) are modified to fulfill the needs of time synchronization in CHWSN Architecture and developed the Cluster Based Hierarchical, Flooding Time Synchronization algorithm (CBH-FTS). It is a hybrid algorithm, where instead of flooding the synchronization messages to neighbor’s node, the root node multicasts the time-sync message to selected cluster- heads using the relevant semantics. Hierarchy of cluster-heads could transmit the synchronization messages down the hierarchy of cluster-heads thus synchronizing only the required part of the network associated with an application. The synchronization could be a result of a decision at root node or could be a result of a request for synchronization from a node to a cluster-head.The CBH-FTS Protocol is semantic driven and covers multiple levels in hierarchy. I. I NTRODUCTION Advances in technology enabled researchers in the field of wireless sensor networks to forecast unprecedented growth of ubiquitous applications.However, anticipated extensive use of these sensor networks has not taken place, partly due to high deployment costs coupled with inflexible designs [1 and 16]. Assumptions such as, homogeneity in sensor hardware and software components, and diversity of application require- ments, H/W limitations need for homogenous network envi- ronments, and highly specific and inextensible designs, have resulted in the inability to accommodate dynamic applications and addition of new applications over the same network [2 and 17]. Designing sensor networks that are deployed once but would respond to ’yet-to-be identified’ applications,dynamically changing network topologies and that scale in multiple di- mensions while adhering to stringent resource constraints is a challenging task. The authors proposed a novel architecture for sensor networks called Deploy Once Multiple Application systems(DOMA) [1]. The DOMA based sensor network ar- chitecture characterized by flexible, semantic driven, multiple hierarchical levels of clusters of nodes is described else where[18 and 19]. The multiple dimensions include multiple modalities of components at the sensor, node, network and application levels that are dispersed over a vast spatially and temporally distributed region. Resource constraints include energy, band- width and processing limitations of each component. DOMA supports multiple applications through the forma- tion of virtual networks of hierarchy of components with varying capabilities and functionalities in a dynamic and autonomous manner. The network components could be added or removed logically and physically from the network. DOMA achieves scalability in a resource efficient manner by reducing information load at each level and in each component of the system through the generation of relevant context driven semantics[1]. Context driven semantics is the ability to utilize semantics associated with a context to affect system behavior[1]. The architecture of the hierarchical, cluster based, flexible network is shown in figure 1. Here more than one application can share the whole or part of the network. Each application may have its own network, processing requirements and pro- tocols. We assumed that whole network is divided into clusters and there is hierarchy of clusters. Fig. 1. Shared hierarchical cluster based wireless sensor network [1]