OVERVIEW

Fluid Infrastructure

Wireless Sensor Network is a network of sensor nodes. A sensor node has, in addition to computation, sensing and wireless communication capability. A network of these nodes is deployed to monitor the environment or detect a target. In the first case, nodes are periodically sampling the environment, and there is a base station that collects the data from them. The data reaches the base station wirelessly by the multihop communication path established by the sensor nodes. The sensor nodes are resource constrained, and communication is the major consumer of battery resource. The nodes near the base station relay other further node’s data, and as a result die earlier. Once they are gone, the network is effectively disconnected. We investigated the use of a rechargeable controlled mobile base station that can traverse the network, collecting data from the nodes. This removes the relaying overhead and increases the network lifetime. The control can be exercised in two domains: time and space. In the first case, we fixed the trajectory & round trip time, and varied the speed to maximize data collection. The mobile base station traverses the same path repeatedly and would dynamically figure out regions of congestion, poor channel etc., and would stop at those places. This was implemented on Berkeley mica2 nodes, and used packbot as the mobile base station. The control in space was formulated as a scheduling problem, was proved to be NP-Complete, and some efficient heuristics were proposed.

Submarine

We introduce a mobile network node to assist with embedded nodes underwater. This underwater data mule is used to carry messages across the network, thereby increasing the longevity of the static nodes. In the some applications, only a few nodes need to keep awake and send the data back to the base station. However, the nodes that are on the path from the wake nodes and the base station have to be awake too to relay the data. If a short cut can be created from the working nodes and the base station, the energy of the other nodes could be saved. One problem in underwater environment is that radio signals attenuate rapidly underwater and long-range underwater communication may not be achieved by using radio. One way to solve this problem is to use sound instead of radio for communication. Another way is to use a messenger, a robotic node that can move itself autonomously through the network. This robot would move from the neighborhood of one node to another node, and exchange messages between itself and the nodes of the network, thus acting as a data mule. The goal of this project is to investigate whether such a robot (a data mule) will reduce energy consumption of an underwater static sensor network and hence increase their battery life.

SYSTEMS / EXPERIMENTS

Fluid Infrastructure

Components in the system

Software Architecture of the system

The experiment was carried out on the topology shown above. The graph below shows the number of packets collected by the mobile in each path traversal.

Submarine

We also developed a distributed algorithm to locate a thermocline underwater (see the Marine Microorganism Monitoring Section), and implemented this algorithm on a test bed. This test bed consists of a linear actuator and 4~5 MICA2 motes. Because of the effect of water, each mote can only communicate with its neighbors over the radio, and hence they compose a simple wireless sensor network. In the experiments to find the thermocline, each node searches a small region to find the potential thermocline and then combines individual search results to estimate where the real thermocline is. The estimation would arrive through a base station. After the first several steps, only a few nodes, named as active nodes, needs to continue their search but lots of nodes still need to stay awake to forward the messages from active nodes to the base station or from the base station to active nodes. The submarine was then introduced to the test bed as a data mule to create a short cut from the active nodes to the base station so that fewer messages would be sent and received throughout the network.

We carried a series of experiments to compare the effect of the introduction of a data mule. The table above shows the number of messages sent and received by nodes in those experiments results. The experiment with the data mule was repeated 3 times and the numbers shown in the table are averages. The last row of the table shows the number of messages broadcasted, which are not counted as the received messages of individual nodes.

	No data mule used		One data mule used
	N sent	N received	N sent	N received
Node 1	9	8	9.33	8
Node 2	4	8	8	11.67
Node 3	14	23	3.67	8.33
Node 4	28	26	18	16.67
Data Mule	N/A	N/A	8	11.33
Base station	10	12	9	11.33
All	N/A	12	N/A	12.33

 

ACCOMPLISHMENTS This Year

The under water data mule was designed and built. Experiments on the data mule with a static sensor network underwater were carried and the results showed that the data mule is able to save the energy consumption of the static nodes.

FUTURE DIRECTIONS

We will continue to explore and refine the fluid infrastructure and submarine developments.

PEOPLE

FACULTY

Fluid Infrastructure:
Prof. Mani Srivastava
Prof. Deborah Estrin

Underwater Datamule:
Prof. David Caron
Prof. Gaurav S. Sukhatme
Prof. Aristides Requicha

STUDENTS

Fluid Infrastructure:
Arun A Somasundara
Aman Kansal
David Jea

Underwater Datamule:
Vitaly Bokser
Bin Zhang
Amit Dhariwal
Eric Shieh