How Smart is your IoT Device?

AUTHOR:  Joyce Li

First off, how do you define smartness? For IOT devices, we can define smartness as how intelligent the device is in predicting future outcomes and trends through big data.  In order to boost how smart your device can be, predictive analytics can help.

Predictive analytics is the practice of extracting information from existing data sets to identify patterns and predict possible outcomes.  Today, in most situations, only 1% of all data is analyzed and used to predict situations.  There is so much potential in data analytics; this 1% is only the tip of the iceberg.

There are two categories in which predictive analytics can be applied to IOT devices: industry-oriented solutions and customer-oriented solutions.1

  • Industry-oriented solutions can help improve costs by:
    • Improving manufacturing efficiency
    • Developing new product genres
    • Reducing warranty costs
    • Reducing recalls
  • Customer-oriented solutions can help companies by:
    • Identifying new customers
    • Identifying new consumer trends
    • Understanding current customers
    • Keeping existing customers happy

There are many challenges to be faced when it comes to processing so much data.  One such challenge is sorting through all the excess data that is generated by the sensor.  Which information is relevant and can be used to predict trends?  A predictive analytics system should have data structures that organize information in the most efficient way, handle time-series based data, filter out data that is irrelevant, and analyze only the important data.

Once sensors are connected to a device, data is continually collected and transferred to the Cloud to be processed in real-time.  The results can be presented on handheld devices such as mobile phones or tablets.   

Here are a few examples of how IoT devices can be smarter:

 

AIR QUALITY SENSORS could be installed throughout China, a country where red alerts due to pollution are regularly issued.  In fact, the U.S. Consulate has sensors installed in Beijing, Shanghai, Guangzhou, and Chengdu, where they measure air pollution throughout the day (seen below).

air-quality-graph

Source: World Air Quality Team Beijing www.aqicn.org

The sensors then issue alerts to the population to warn them of potentially poor air quality.  A red alert signifies that the Air Quality Index (AQI) is between 50 and 200, which is considered unhealthy.  AQI levels in Beijing have far surpassed that limit, reaching 611 in Beijing last November.  Air pollution causes about 17% of the deaths in all of China.

With predictive analytics, we can make measuring air quality smarter.  Rather than just collecting air pollution data, we could predict which days would be unhealthy, very unhealthy, and hazardous, allowing people in China to plan their outdoor activities accordingly.  By analyzing the data, the quality of life could be drastically improved in China.  Sensors would pick up potential problems in air quality, trends would be analyzed and citizens would know in advance of possible deteriorating air quality before it reaches a dangerous level.

 

BAROMETRIC SENSORS could be used to predict hurricanes that strike many parts of the world.  Hurricanes are categorized 1-5 on the Saffir-Simpson Hurricane Scale; category 1 being mild and category 5 being catastrophic.  Any hurricane between category 3-5 would be considered dangerous for the general public.  Hurricanes typically occur in regions of low air pressure; more specifically, hurricanes from categories 3-5 have pressure levels 28.49 in Hg and below (normal pressure is 29.29 Hg).

Currently, many cities that live in threat of hurricanes do try to detect and predict hurricanes.  Using a Doppler radar, cities use rainfall levels and patterns to determine the shape and direction of the hurricane.  Scientists also use simple computer programs to take information from satellites and radar to predict the path of the hurricane.  When a hurricane is detected, news is broadcast to citizens through radio stations, TV programs, and more.  However, currently, the predictions are not very accurate and have errors up to 100 miles; that distance could mean the difference between evacuating one city or another.

Using a smart barometric sensor, hurricanes could be predicted more accurately ahead of time giving citizens early warning of where they will strike.  Measuring air pressure instead of rainfall could yield more informative results; the lower the pressure, the bigger the storm.  Using sensors placed in hurricane-prone areas, we could generate alerts based on big data analysis of weather data and predict hurricanes with higher accuracy.  

hurricane-photo

Photo courtesy of The Old Farmer’s Almanac

 

TEMPERATURE and MOTION sensors can be applicable in factory settings. Within factories are many machines that could possibly be dangerous if they overheated or malfunctioned.  With a temperature sensor, overheating could be prevented.  A notification would be sent to the operator of the machine if the temperature was steadily increasing out if its normal range.  Meanwhile, a motion sensor could also be used to monitor the machine’s speed to measure its efficiency, as well as caution the owner if the speeds are dangerously high or low. Over time, with a database of machine data, you could predict how often a machine would malfunction, thus preventing potential accidents in the future.

factory-photo

Photo courtesy of CBInsights

Ultimately, these sensors can be applied on a larger scale, predicting air quality crises and machine malfunctions and making many of the world’s cities smarter,  but they could also be attached to any IOT device and help people in their everyday lives.

 

References:

  1. http://www.predictiveanalyticsworld.com/patimes/opportunities-and-challenges-predictive-analytics-for-iot/7775/?utm_medium=email&utm_campaign=PATIMES&utm_source=Patimes-list&utm_content=07-12-16-July-Edition)

Author: Rochelle Drenan