Efficient communications and data compression algorithms in sensor networks using discrete wavelet transform

• US 7,715,928 B1
• Filed: 05/17/2005
• Issued: 05/11/2010
• Est. Priority Date: 05/17/2004
• Status: Expired due to Fees

• Current Assignee: University of Southern California
• Original Assignee: University of Southern California
• Inventors: Ciancio, Alexandre, Ortega, Antonio
• Primary Examiner(s): Patel; Ramesh B
• Assistant Examiner(s): GAMI, TEJAL
• Application Number:
  US11/131,916
• Time in Patent Office: 1,820 Days
• Field of Search: 348/88, 382/240, 700/1, 340/539.19, 702/189, 375/240.12
• US Class Current: 700/1
• CPC Class Codes: H04L 67/12 specially adapted for proprietary or special-purpose networking environments, e.g. medical networks, sensor networks, networks in vehicles or remote metering networks
• CPC Class Codes: H04L 67/5651 Reducing the amount or size of exchanged application data

Abstract

Techniques and sensor networks based on distributed wavelet transform algorithms for efficient sensor networks. Wireless multihop sensor networks can implement the described techniques to exploit the data flow direction in the network to perform partial computations to approximate the wavelet coefficients using the available data that arrives at each sensor. Non-multihop sensor networks may use inter-sensor communications and the distributed wavelet transform algorithms to decorrelate the data and thus improve the energy efficiency.

Claims

1. A sensor network, comprising:

a plurality of sensors distributed in an area to acquire data of the area;

communication links to link at least a portion of the sensors to allow for communications amongst linked sensors;

a central node in communication with the sensors to receive sensor data from the sensors;

with each linked sensor comprising a data processing mechanism to perform a distributed wavelet transform of data collected by the each linked sensor and received from neighboring linked sensors to produce transformed data for the each linked sensor which has less data bits than the data collected by the each linked sensor, and with the each linked sensor transmitting the transformed data to the central node in lieu of the data collected by the each linked sensor.

2. The sensor network as in claim 1, with at least one sensor located near the central node transmitting data that is collected by the at least one sensor and does not include data from other sensors to the central node without performing the distributed wavelet transform.

3. The sensor network as in claim 1, further comprising:

second communication links to link a second portion of the sensors, with each linked sensor in the second portion of the sensors comprising a second data processing mechanism to perform a second, different distributed wavelet transform of data collected by the each linked sensor and received from neighboring linked sensors in the second portion to produce transformed data for the each linked sensor which has less data bits than the data collected by the each linked sensor, and with the each linked sensor transmitting the transformed data to the central node in lieu of the data collected by the each linked sensor in the second portion.

4. The sensor network as in claim 1, with the linked sensors including a plurality of sensor clusters where each sensor cluster comprises a plurality of linked sensors, the sensor network further comprising a digital signal processing device which comprises:

an input to receive an input signal having a plurality of samples received from different linked sensors; and

a digital filtering device having a plurality of filtering stages to perform a discrete wavelet transform on samples received from said input to produce data representing a discrete wavelet transform of said input signal, said digital filtering device operable to perform said discrete wavelet transform on one block of samples from linked sensors in one sensor cluster, with said digital filtering device completing computations that are computable based on information solely from said block and perform partial computations on samples in said block which requires information from a different block for another sensor cluster immediately preceding said block, without information from said another block.

5. The sensor network as in claim 4, with the digital signal processing device further comprising:

a memory unit with a memory space sufficient to store said information from said another block, having an input terminal coupled to receive data from an output of said digital filtering device and an output terminal to supply said information to said digital filtering device, with said digital filtering device receiving said information, after completing said computations and partial computations on said block, to complete remaining operations of said partial computations.

6. The sensor network as in claim 1, with the communication links comprising wireless links.