Michelle Kelsey, PLM, Sensor Solutions Division, Freescale, Inc.


“Big Data” is here, but it is still in its adolescence and its growth is expected to continue. One of the main sources of big data may be the smallest components.  Sensors may be perceived to generate “little data”, but it adds up quickly. As Walmart claims to have access to “petabytes” of consumer data from 145 million Americans (more than 60 percent of U.S. adults), sensors are providing factors of 100 times that transaction data for the same consumers who are walking down the shopping aisles.  Unlike the data that may result in each customers purchase transaction, multiple sensors will autonomously generate multiple measurements of data every few seconds for the same person.   As sensors are becoming smaller, we can already see multiple sensing technologies becoming more prominent in mobile products, automobiles, smart cities, smart farms and smart appliances. As the Internet of Things (IoT) continues to grow, sensor data will become the largest source of “Big Data”.

From only a person’s mobile phone, sensors will detect the last accurate GPS location, the relative position since that last accurate GPS position, the direction the person is facing, how bright the location is where the person is standing, what the ambient pressure and humidity may be where the person is standing, how many steps the person has taken, how much pressure the person’s feet experienced from each step, (even determining how many more steps until new shoes are recommended) and the person’s cadence, speed and calories burned. All this happens during that same hour. That is how quickly sensor data can become the new big data.  This is not the future, but rather what is happening now using just the typical sensors that already exist in your phone such as an accelerometer, gyroscope, magnetometer, pressure sensor, GPS and WiFi. We are in the advent of the wearables adoption as iSuppli predicts there will be 92.5 million wearable devices in 2016. Therefore, more people will be wearing smart watches and bands with these sensors. There will be sensors in our shoes, and eventually they will be integrated into our clothing.  At the same time, Smart Cities are predicted to increase four times (iSuppli) from 2013 to 2025, using sensors to manage water distribution, electric tampering, electric car infrastructure, traffic control as just a few examples of compilations of sensor data. The sensor data is adding up.


Most consumers have at least five but as many as ten unique sensor types on them at all times.  In most smart phones you will find a proximity sensor monitoring how close the phone is to your face to determine when to turn off the display. An accelerometer detects orientation, motion and tapping to determine the display for landscape and portrait mode, or provide tap selection or motion control for your user interface. An ambient light sensor detects the amount of light that is available and will adjust the brightness of the screen to conserve battery power. A moisture sensor in the dock connector determines when you may have dropped your phone in water and a gyroscope measures fast rotations for games and when used with an accelerometer and magnetometer, it can provide accurate relative position determination. That is just in your phone. If you carry a tablet, you have just doubled your sensor data.

When driving in your car, you have increased the sensors you interact with by a factor of three. Aligning to Advanced Driver Assistance Systems (ADAS), your car comes standard with accelerometers to determine a front side crash and gyroscopes to provide electronics stability control. Each tire in your car has a tire pressure monitor sensor (TPMS) that includes a combo sensor that includes pressure sensing, temperature sensing, acceleration sensing and battery sensing to determine when a tire is moving, when pressure changes occur, and if temperature change was the reason for the pressure change or if it was due to actual air pressure reduction by volume.  Your next car will include even more sensors as the amount of sensors increases with a focus of sensing the driver and the conditions in the cabin.

Smart cities will utilize sensors for everything from monitoring physical infrastructure such as bridges and roads, to people. Embedded sensors will measure vibration, traffic and linear displacement (cracks in the cement). There are 600,000 bridges within the entire United States highway network. USA Today reported that over 65,000 US bridges are in need of repair. If 100 accelerometers are used per bridge to monitor traffic, motion, cracking and other disturbances, then 6.5 million sensors would be transmitting data on just the US bridges requiring upgrade and repair. The US highway network alone consists of four million miles of roads and streets. If we consider the number of global bridges, highways and roads, there will be a great deal of sensor data adding up.


One dimension of IoT has spawned by the perpetual technical improvements of a different set of applications.  Wearable sensors used to accommodate the need to quantify everything continuous to open new markets with focus on the ultimate wearable solutions that are virtually invisible to the user. Sensor products are now in jewelry and eventually will be seamless and interweaved into clothing. With chip scale packaging technology utilizing 3D interconnection approaches such Thru Silicon Via (TSVs) technology, inertial sensors are able to achieve 77 percent smaller footprint area than five years ago while still increasing sensor functionality with higher integration density.  With these integration technologies in MEMS solutions, integration has scaled from 3-axis to 6- and 9-axis sensing to reduce the overall footprint over three times, and power has decreased over 80 percent from three years ago (gyroscopes alone have decreased power consumption by 50 percent, from 5mA full operating power to 2.6mA with products such as Freescale’s FXAS21002 3-axis discrete gyroscope within one year). Sensor size and power reduction to date have enabled adoption for mobile phones, tablets, smart watches, homes and appliances, but the move to wearables and IoT will be the main drivers to the next generations of even lower power and size. An alternate dimension continuing the technology innovation is in automobiles as the there is more focus on looking into the cabin and to include driverless vehicles. According to ABI Research, Global Driverless Vehicle Shipments are projected to Reach 14 million by 2030. Therefore, there is a need for higher sensor integration and sensor analytics to make decisions based on the information the sensors are providing.


To meet the demands of mobile products, sensors have evolved to become smaller, but the requirements for wearable products and IoT have put even higher expectations of smaller form factor, with much lower power and with higher processing on the ASIC.  At the same time, the sensor data needs to be managed at a local level (sensor hub) and corrected and compensated (Sensor Fusion). A Sensor Hub is essentially a manager of the various sensor types or in the case of combo devices, the sensor operation.   Sensor fusion is a process by which data from several different sensors are “fused” to compute something more than could be determined by any one sensor alone.  Basically, each sensor has its own strengths and weaknesses but each can be calibrated by another sensor to remove drift or identify environmental disturbances.

Sensor Fusion enables complementary sensors to compensate for another sensor weaknesses, providing a more accurate overall solution. For example, a gyroscope’s drift over time is measured and modeled by an accelerometer’s lack of motion and removed by the sensor fusion before the data is processed. A magnetometer measuring Earth’s magnetic field for heading information can be affected by a magnetic field from a speaker or a tablet or phone holster magnetic, but sensor fusion algorithms such as can demonstrated by Freescale’s Sensor Fusion app, can be removed as shown by the calibration and compensation as the accelerometer or gyro did not sense any physical motion.  With all these technologies, sensors are able to be easily integrated into smaller battery powered solutions without any compromise on accuracy.


Not all the sensor data will be considered Big Data, but it will be the job of every level of the solution to manage the data, determine what data should be secured, what sensor data has business potential and what data are being more enveloped in our daily lives.

The use case of the sensor data will determine what level of data is actually transmitted to the cloud from raw sensor data,  calibrated and compensated (Sensor Fused) Data or physical data (i.e. steps taken, speed, gestures detected).

The determination of the data that is transmitted to the cloud will be the cost of intelligence at the physical nodes and more heavily weighed by the power available at the physical nodes. Although processing data locally would allow for less data transmitted to the cloud, it would require higher processing power and therefore, higher costs and power requirements at each sensor node. Maintaining the raw sensor data could have some benefits however, such as allowing updated algorithms to be used on the same foundational sensing data for higher accuracy results.  Therefore, data reduction will be limited until there are adopted sensor frameworks defined for the smart cities and mobile device operating systems that require it. It is too soon to tell, but in most cases, data redundancy will be expected.


Another determination of the amount of sensor data that will be transmitted is the acceptance to share personal sensor data or purchase the products that will share personal sensor data. A person’s acceptance, level of security, liability issues, and regulation will continue to challenge concerns over the decisions of who and what should have access to this data. Your personal data, your driving information, your home monitoring, the roads you drive on and the appliances you use. Smaller concerns may be around product warranties that may be nullified depending on how it is handled and in place of the standard one to ten year warranty limit. We see some of this today with the water sensor in your phone. Larger concerns are worried about your car dispensing automatic tickets when going over the designated speed.  These may be some initial thoughts.  However, sensor data that may initially not be considered relevant or desired can be very relevant to your time, your environment, your security and most importantly, your safety.

Drivers may not want to share their driving behavior with another user’s vehicle. But if the benefits can be well understood, the adoption rate would increase significantly.  If your vehicle could receive information that the car behind you typically passes on the right and will be turning left soon as it is planning on doing from the planned route it is following, you could be guided with the best way to manage that car. If your vehicle could share its sensor data, it could allow for more efficient management of traffic, which in turn reduces commuters’ driving time on the road and reduces the gas used while stuck in traffic, but more importantly reduces the probability of accidents. According to the annual Traffic Scorecard compiled by the traffic information and driver services provider INRIX in Kirkland, WA, in 2013, U.S. travelers spent an average 47 hours sitting in traffic. That time and gas could have been reduced if the driver was notified of the traffic times and left a little earlier or later, or a traffic accident was avoided by smart radar or car acceleration or deceleration changes.  According to the Association for Safe International Road Travel, in 2013 over 37,000 people died in road crashes each year in the U.S. (nearly 1.3 million people globally) and 2.35 million (20-50 million globally) are injured or disabled.  Sensor data can help to measure the speeds of other cars or warn of cars that are not easily seen by the driver.  Additionally, sensor data that is transmitted to other cars can allow for more educated driving decisions. For example, if a car is coming to an intersection, you can receive a signal that the car has driven through stop signs 15 percent of the time.  These solutions can help drivers today, but more importantly in the future.  Ford estimates the number of automobiles on the world’s roads will rise from around one billion today to as many as four billion by 2050. Therefore, there is an increasing need to manage the traffic, inform users of the safest and fastest direction to drive and to reduce the carbon footprint.

Another important area is in farming.  Precision Agriculture (or Precision Farming) uses Global Positioning Monitors (GPS) with wireless communications to provide detailed sensor data for the farmer to monitor the environment, track crop growth, to make predictions of crops expected but more importantly, enhance farming techniques, increasing efficient and effective farming processes.


With ‘always on’ autonomous data there is a need for securing the data and only sending the information that is granted for access to certain applications or users. This security is continuously updated and built into the OS of your mobile products Examples of this are with Android Wear, a version of Google’s Android operating system designed for smartwatches and other wearables. New markets are seeing new ways to communicate to their smart products such as Thread or the open source AllJoyn protocol that provides simple ways to interact with their nearby things. Security will be needed to be built into the buildings, structures, cement and automobiles.  These discussions are happening at events and consortiums that focus on setting current and future policies to guarantee public security.  Smart sensor solutions for urban cities will be adopted more easily with ensured highly secured data.   Freescale Intelligent Sensing Framework will allow customers to go to market quickly with their solutions. They allow for developers to focus on the application of the raw sensor data into their solutions. This framework can be open source, but the processed sensor data and application data will need to be secured to the cloud during processing. Moving forward, more sensor analytic capabilities will be embedded into the solutions to offer another layer of intelligence to next generation products.


Sensor data is already a big component of “Big Data”. Sensors generate the “little data”, but the data can help predict earthquakes, structure weaknesses in buildings and bridges. The “petabytes” of consumer data may help provide the latest sales and push ads to consumers for increasing sales.  But the ‘always on’, autonomous sensor data will generate multiple measurements of data to provide more safety and environmental changes that will be prominent in mobile products, cars, smart cities and smart appliances. As the IoT continues to grow, sensor data will become the largest source of “Big Data”, managing the data will be a challenge for some implementations, but if done efficiently, it can present success in business opportunities. More importantly to the consumers, it enables success in other less prominent business areas such as –  efficient driving and parking, safer commutes, protection of our loved ones with monitoring devices for our children, elderly and pets, more secure home monitoring, safer bridges, safer buildings, less water used, more efficient crops, and more efficient machines. With increasing levels of sensor data, it will be more reliant on the sensor analytics behind it. As with all “Big Data” content, it’s value will be dependent on the insight and information that can be abstracted from the data.