An Energy-Efficient Data An Energy-Efficient Data Storage Scheme for Multi-Storage Scheme for Multi-

resolution Query in Wireless resolution Query in Wireless Sensor NetworksSensor Networks

老師 : 溫志煜學生 : 官其瑩

Sensor Sensor

Sensor is a very small device with its own processor, memory and battery. Sensor node sensing data, then storage and the transmission signal.

Power saving is the problem which many people studied :

We observed that there are high relations between the data storage method and the power consumption. Wasting too much power is inefficiency while transmitting, and it will reduce the network lifetime. We should deal with this problem when we design the storage scheme.

main propose of power saving mechanism: let the sensor node turn into sleep mode to preserve power.

What if the data user queried is in the sleeping sensor node?

Does sensor nodes remove its data before it turn into sleeping mode?

How about the sensor nodes awoke?

Three basic data storage schemes used in wireless sensor network:

1. External Storage (ES) 2. Local Storage (LS) 3. Data-Centric Storage (DCS)

In this paper, we proposed a data storage scheme to arrange sensor node’s sleeping schedule and acquire multiple resolutions of sensing data in wireless sensor network.

The basic idea of our scheme is to divide network into many grids. Each grid has its own sleeping schedule which sensor nodes guide this to turn into sleep mode.

The transmission range of sensor node and the grid’s length :

r : The transmission range

d : length of a grid

Arrange sensor nodes into rows and columns and fill the matrix with time slots:

The time slots distribution matrix:

sensors : A, B, C, D, E, F,G

time slots:1~12

Arrangement of wake up intervals of the time slots distribution:

The properties of snake liked time slots distribution

Property 1: There are only two sensors assigned in the same slot in the matrix.

Property 2: There must be a continuous time slot assignment overlapping two sensor’s wake up period.

Property 3: The beginning time slot and the last slot of matrix will engage to complete a cycle.

Property 4: The time slots distribute matrix has limited fault tolerance ability

Store the Event

As sensor node sensed event, it will hash this event key to decide where to store. Then the sensor will send the data to destination grid. We apply a GPSR liked scheme to rout. While the data transmit to the destination grid, it will store in the two sensors which wake up in this time.

Store data in the corresponding grid with two wake up sensor nodes

Data Precision

The snake like time slots distribution scheme’s special structure lead to a new view of the multi-resolution. The idea is to calculate the time slots proportion which it got of all time slots in the matrix. We assume every time slot have the same probability to store data. The matrix can be regarded as all of the data.

The portion of data that queries time slot 5

Retrieve the Events

The query process in our data storage scheme will face the delay problem, because the data precision is operated in an accumulating manner. The location that query is interested can be hashed by the event key.

As the query arrives in the destination grid, the wake up nodes can return its data if the precision is satisfied. Otherwise the query data will be held to wait for another nodes’ data to increase the precision.

CONCLUSION :In this paper, we have proposed an energy- efficient storage scheme that enhances the data reliability. The snake liked time slots distribution arranges only two sensors wake up in a grid. Therefore, the rest of the sensors can turn into sleeping mode. The data will be stored in wake up nodes by a DCS manner. As a node hashing the event to decides where to be stored, it will be transmitted to corresponding grid.

In addition, sensor node’s wake up period will be overlapped by other sensor’s wake up period. The data will not be lost when sensors turn into sleeping mode. Because there must be a sensor wakes up in the same time and takes care of the data.