Query Dissemination and Routing

This process will start after the query optimization phase. It means the optimized queries must be propagated throughout the network. So this can be started from the root node of the network with a broadcast of the query. In this case, the query must notice to each sensor in the network based on the query applied method. If the query applies locally then it must be broadcast to its child nodes also in the network routing tree. This is because in Aquisitional query processing technique it is important to identify the path of query where to be disseminated. According to this concern, it will help to reduce the costs of queries incorrect initiating in Acquisitional query processing environments such as in TinyDB application.

When doing query dissemination process in the network that there are two important concerns should be taken into account. Those are, if a query does not affect at an exacting node, and the node that is query applied doesn’t have any child nodes, then there can be removed the whole sub tree from that query. In doing so, it is much more conserve battery power of the sensor nodes reducing the costs of query dissemination, executing and results forwarding across the network. As well as, it will concern about nodes of the network with extending the life time of them. In this case, the process is to determine the node that need not participate in a particular query is a difficult task. But there are two situations have been considered to be taken into account to make this task in better way.

One usual situation is having constant valued attributes with a selection predicate that is nodeid or location. It shows that those nodes do not need participate for the query. Likewise, if a node knows that none of its child nodes will ever satisfy the value of some selection predicates then it does not also need to forward the query to its sub tree. It seems to be an energy efficient way in query dissemination process this is because, before a particular query disseminates it searches the query path that should be queried upon conditions earlier mentioned. In this way, making the routing information of the network in a better way and that will benefit for sensors to extend their lifetime much more.

Semantic Routing Tree

The primary goal in here is to determine whether it needs to participate or not below child nodes in a given query over some constant attribute value. In generally says, an SRT is a distributed index to locate relevant nodes in the network. Usually, each node stores value interval of itself and its children also in the network. Therefore this will help for SRT to determine a particular node that is needed to participate or not in a given query. In this mechanism, it may be important task when making the selection for the parent node to implement the SRT routing tree algorithm. Traditionally, when doing the routing tree algorithm construction is done by having nodes pick a parent with the most reliable connection to the root in sensor networks. But in this case it is differ with respect to the SRT. Therefore, in making the selection the parent in an SRT is considered such that several parents of comparable link quality. This process is known as link-quality-based parent selection algorithm in the SRT. It will give more details about link-quality-based algorithm on [Woo A. and Culler D. July 2001].

In this case, theoretically the SRT can take as an index over constant attribute value of a sensor node. That is used to locate nodes that have relevant data with respect to the query. The important thing is in here that the SRT is an overlay on the network. This is because SRT keeps all indices of sensor nodes with regard to the constant attribute value. As well as each node in the network maintains a single one-dimensional interval showing the range with regard to the constant attribute values underneath every child nodes of the network tree. Usually, this will help to make efficient routing process throughout the network before particular query dissemination. The query processing task is proceeding in the network such that when a query q (as an example) with a predicate value say A (as an example) arrives at node n (as an example), then node n check whether if any child’s node attribute value of A overlaps the range of A attribute value in query q. If this happens, it receives result and forwards the query and also if it is not happened the query is not forwarded. As well as, if the query applies locally then node n begins to execute the query itself [Madden et al., 2003].