Supervisor: Dr Weifa Liang
E-Mail: wliang@cs.anu.edu.au
Background
Wireless sensor networks have received significant recent attention due to their potential applications from civil to military domains including military and civilian surveillance, fine-grain monitoring of natural habitats, measuring variations in local salinity levels in riparian environment, etc. The sensors in sensor networks periodically collect the sensed data as they monitor their vicinities for a specific interest.
The sensed data can be transmitted directly to the base station where the client queries are posted, the base station then processes the collected data centrally and responds to users queries. However, due to the fact that each sensor is equipped with energy-limited battery, the energy efficiency rather than the response time for each query is paramount of importance in sensor networks. The energy consumption for the above centralized processing is very expensive. Instead, we adopt energy-efficient in-network processing, i.e., the data gathering is implemented through a tree rooted at the base station, and the data at each sensor has to be filtered or combined locally before it is transmitted to its parent.
Unlike most existing assumptions that each sensor only transmits a fixed length of message, we consider another data gathering, in which the length of the message that each sensor transmits to its parent is proportional to the sum of the message lengths of its children and itself. Thus, the total energy consumption at a node is proportional to the length of the transmitted message and the distance between the node and its parent in the tree. Specifically, on one hand, we aim to find such an energy efficient spanning tree rooted at the sink node that the network lifetime is maximized. On the other hand, we prefer the approximate result instead of the exact result in traditional data models for a given query, since the communications among the sensors are fail-prone. Under this circumstance, how close the approximate result is to the exact result becomes the essential issue.
In summary, this project will be in pursue of energy efficiency on the basis of database modeling in sensor network environments. A student working on this project would be involved:
Supervisor: Dr Weifa Liang
E-Mail: wliang@cs.anu.edu.au
Recent advances in microelectronic technology have made it possible to construct compact and inexpensive wireless sensors. Networks formed by such sensors, termed as wireless sensor networks, have been receiving significant attention due to their potential applications in environmental monitoring, surveillance, military operations and other domains. While these applications revealed tremendous potential for capturing important environment phenomenon using sensor networks, they have also posed certain associated limitations. One of the major limitations of sensor nodes that they are powered by low power batteries, which limit the network lifetime and impact on the quality of the network. Energy conversation in sensor network operations is thus of paramount importance. As most currently deployed sensor networks are used for monitoring and surveillance purposes, the sensed data generated by the sensor network must be available for users at the base station, data gathering that extracts and collects sensed data from sensors thus is one fundamental operation in sensor networks. The energy conservation on data gathering leads to prolongation of network lifetime. The study of energy-efficient data gathering poses great challenges due to the unique characteristics and physical constraints imposed on sensor networks. In this project we study the {\it aggregate node placement problem for data gathering}, which aims to place a few powerful aggregate nodes to a dense sensor network such that the network performance can be significantly improved. Assume that a heterogeneous sensor network consists of two types of sensor nodes: cheap sensor nodes that have fixed identical transmission range and expensive aggregate sensor nodes that have various transmission ranges. Aggregate nodes have more energies, higher data transmission rate, and better processing and storage capabilities than sensor nodes. The main task of an aggregate node is to process and relay data for sensor nodes and the other aggregate nodes. Due to expensive cost associated with aggregate nodes, the number of them in a sensor network is very limited, thus, they need to be placed in the network carefully in order to increase the network connectivity, reduce the data gathering delay, and prolong the network lifetime. Specifically, the aggregate node placement problem for data gathering is as follows. Given a sensor network consisting of $n$ dense sensor nodes that are randomly deployed in a region of interest and a few number of expensive aggregate nodes $K$ ($K<< n$), the problem is to place the $K$ aggregate nodes in the region so that the network lifetime can be further maximised by answering data gathering queries.
Supervisor: Dr Weifa Liang
E-Mail: wliang@cs.anu.edu.au
Background
Technological advances in recent years have enabled the deployment of large-scalable wireless sensor networks (WSNs) consisting of hundreds or thousands of inexpensive sensors in an ad-hoc fashion for a variety of environmental monitoring and surveillance purposes. In these applications, a large volume of continuous sensed data generated by sensors needs to be either collected at the base station or aggregated within the network. Thus, WSNs usually are treated as a virtual database. The skyline query, as an important operator in modern databases for multi-preference analysis and decision making, has received much attention recently due to its wide applications. In this project we consider the skyline query problem in WSNs.
We aim to devise novel, distributed evaluation algorithms for skyline query evaluation from the scratch. We also consider the skyline maintenance within sliding window environments by developing distributed maintenance algorithms for it. We will conduct extensive experiments by simulation to evaluate the performance of the proposed algorithms in terms of multiple performance metrics including the total energy consumption, the maximum energy consumption among the network nodes, and the network lifetime.
A student working on this project would be involved: