System and method for determining information and outliers from sensor data

A system for determining information from sensor data, the system comprises:

• a processor;
• a memory coupled to the processor, wherein the processor is capable of executing programmed instructions stored in the memory to:
• receive sensor data provided by a sensor, wherein the sensor data comprise periodic data, random data, or non-stationary data, wherein the periodic data include Electrocardiogram (ECG) data and Photoplethysmogram (PPG) data, wherein the random data include Electroencephalogram (EEG) data and smart meter data and wherein the non-stationary data include the smart meter data, the EEG data, stock data, and accelerometer data;
• determine statistical features of the sensor data using data processing techniques;
• determine signal dynamics of the sensor data based on the statistical features, signal processing features, and a data distribution model, wherein the data distribution model comprises Autoregressive Moving Average (ARMA) based model or Autoregressive Integrated Moving Average (ARIMA) based model and wherein the signal processing features comprises features associated with periodicity;
• identify at least one outlier class based on the signal dynamics and number of streams of the sensor data, wherein the outlier class is at least one of point detection, collaborative, or contextual and wherein the point detection outlier class is related to one of a distance parameter or a density parameter;
• select at least one outlier detection method for detecting outliers in the sensor data based on the identified at least one outlier class;
• determine information content of the sensor data based on detected outliers, the signal dynamics, the statistical features, and information theoretic features of the sensor data, wherein the information content of the sensor data is a measure of one or more features of the sensor data and is determined using a data modeling technique corresponding to the measure of one or more features of the sensor data and wherein the information theoretic features indicate variation of information and entropy;
• determine an information score for the information content of the sensor data based on the signal dynamics, the identified at least one outlier class, wherein the information score is indicative of amount of information including a measure of similarity or dissimilarity score associated with each outlier;
• normalize the information score by removing identifying outlier errors, wherein the identifying outlier errors are one of false positives;
• identify one or more outliers with highest information content based on the information score and
• validate the outliers using adaptive data compression and adaptive dissemination and validate the information content of the sensor data based on the signal dynamics.

A non-transitory computer readable medium embodying a program executable in a system for determining information from sensor data, the program comprising:

• a program code for receiving sensor data provided by a sensor, wherein the sensor data comprise periodic data, random data, or non-stationary data, wherein the periodic data include Electrocardiogram (ECG) data and Photoplethysmogram (PPG) data, wherein the random data include Electroencephalogram (EEG) data and smart meter data and wherein the non-stationary data include the smart meter data, the EEG data, stock data, and accelerometer data;
• a program code for determining statistical features of the sensor data using data processing techniques;
• a program code for determining signal dynamics of the sensor data based on the statistical features, signal processing features, and a data distribution model, wherein the data distribution model comprises Autoregressive Moving Average (ARMA) based model and or Autoregressive Integrated Moving Average (ARIMA) based model and wherein the signal processing features comprises features associated with periodicity;
• a program code for identifying at least one outlier class based on the signal dynamics and number of streams of the sensor data, wherein the outlier class is at least one of point detection, collaborative, or contextual and wherein the point detection outlier class is related to one of a distance parameter or a density parameter;
• a program code for selecting at least one outlier detection method for detecting outliers in the sensor data based on the identified at least one outlier class;
• a program code for determining information content of the sensor data based on detected outliers, the signal dynamics, the statistical features, and information theoretic features of the sensor data, wherein the information content of the sensor data is a measure of one or more features of the sensor data and is determined using a data modeling technique corresponding to the measure of one or more features of the sensor data and wherein the information theoretic features indicate variation of information and entropy;
• a program code for determining an information score for the information content of the sensor data based on the signal dynamics, the identified at least one outlier class, wherein the information score is indicative of amount of information including a measure of similarity or dissimilarity score associated with each outlier;
• a program code for normalizing the information score by removing identifying outlier errors, wherein the identifying outlier errors are one of false positives;
• a program code for identifying one or more outliers with highest information content based on the information score and a program code for validating the outliers using adaptive data compression and adaptive dissemination and validating the information content of the sensor data based on the signal dynamics.