WSN Data Aggregation - Giuseppe Visalli Personal WebPage - Italy

Go to content

Main menu:

WSN Data Aggregation

Inventions
A Ultra-Low Energy Data Aggregation is presented for wireless clustered sensor network. The proposed approach optimizes the energy budget in the network, composed by several leaf cells (sensors), a plurality of cluster head and one base station. The aggregation function compresses digital data, after analog to digital conversion, by a binary operator used in the coding tecnique found here. The "S" operator, scrambles the binary discrete time random process lines by a permutation pattern. Each pattern has an unique inverse that allows the data recovery. The aggregation function is the bit wise XOR of permutated data by a periodic sequence of permutation patterns called "the signature". Data recovery is  a process of dis-aggregation, probabilistic optimization and low-pass filtering. The dis-aggregation is the application of the inverted signature to the compressed binary beam, resulting the source sensor data corrupted by a detection noise. The probabilistic optimization is a profile of the detection noise based on the measurement of the joint probability of the noise and a record of the aggregated beam. This step reduces the noise power resulting a final random noise with probability mass function assimilabe to a Gaussian distribution. Each line of sensor binary data is profiled with a delay line for the aggregated beam; to reduce hardware complexity, but decreasing the efficacy of the approach, the most significat bit of the process uses a deep delay line, instead the less significant bit has a lower accuracy. Since this different approach, resulting data is assimilable to the original sensor data corrupted by additive high-frequency noise. Low Pass filtering recover most of the sensor wave. We defined the E% the energy relative error, comparing the energy of data from sensor and the wave recovered at the base station. Experimental results show that E% is at most 10% when sensor process is auto regressive with relative bandwidth of 0.01 and a data frame of D=2048 bytes.
Dis-Aggregation, Optimization and Low-Pass Filtering (LPF)
A non trivial example
Gaussian, Saturation, Rayleigh and Exponential Decay
A delay is introduced since phase spectrum of low-pass filter
 
Back to content | Back to main menu