The German philosopher Immanuel Kant described space and time as a priori notions that enable us to comprehend “sense experience” and contextualize the world around us. What he was talking about is how people make sense of the physical world. Most of us use a combination of instinct mixed with maps, clocks, visual cues, and reasoning. While effective, such practices still leave us with a somewhat limited and skewed perspective. After all, we can consume only so much data about our physical surroundings and reason only so fast.
Machines obviously can process data at extremely fast rates, but how do we provide machines the necessary context to approximate this “sense experience” described by Kant? Machines experience space and time in a much more explicit way than we do. Maps and visual cues don’t work well for machine consumption. Machines require a complete, precise, and real-time digital representation of space and time. While such a perspective may not match our own, it is critical to machines’ having a functional model of their environment. We call this contextual awareness.
Contextual awareness is rapidly becoming the currency of the 21st century. While you hear a lot of talk about big data, the fact remains that big data has not been about real time, not until now anyway. The majority of big data projects have been focused on batch processing large volumes of data, much of which was generated on the Web. Contextual awareness takes big data one huge step forward.
The convergence of billions of machine-generated data points per second, contextualized by spatial data, will enable the emergence of autonomous things and will have a profound impact on businesses. We will see entirely new business models emerging, blurring the line between traditional/human/physical awareness with digital/virtual/autonomous ones. It’s less a question of “if” and more a question of “how rapidly” it will happen. To illustrate my point, I offer several real-world examples of contextual awareness and the potential such advancements hold for people and businesses.
The Connected Cow
About 15 years ago, a friend of mine ran a start-up company that put radiofrequency identification (RFID) tags on cattle. What he had helped design was the agricultural equivalent of a factory inventory control system. Anytime a tagged cow walked past a fixed receiver, the bovines’ unique identifier, along with a time stamp and reader location, would be recorded.
Such automated technology represented a significant advancement in the management of large herds. It was a way to say goodbye to what had been a tedious pencil and paper or keyboard entry operation. Yet, for all its positive attributes, the system captured only a limited amount of data, from a few fixed locations, at a cow’s-pace frequency, resulting in just a few records per day. These recordings were saved on a local disk, physically moved to an office once a week, uploaded to a desktop computer, and then consolidated into a spreadsheet program. Some of what you gained in the convenience of replacing a manual process, you lost in cost on a relatively expensive sensor (for its time), a limited communication network, and minimal processing capability.
Enter the Connected Cow. As Cisco detailed in this infographic a few years back, the Dutch company Sparked developed sophisticated sensors that cattle farmers could implant in a cow’s ear. The data from the cows, transmitted wirelessly to the farmer, could be used to determine the precise location and health of individual cows and entire herds. Farmers also could learn about fluctuations in diet, how cows respond to environmental factors, and general herd behavior. Each Connected Cow transmits approximately 200 MB of data per year. For a very large dairy farm, this technology could generate more than 3 TB of data annually.
Looking ahead, companies such as Vital Herd are preparing to introduce additional sensors that will be ingested by the cows to provide detailed information about practically every biological parameter imaginable. These sensors will generate this information at a pace much faster than humans can consume.
What does this mean to the cattle farmer? In the not too distant future, the Connected Cow will help diagnose health issues and enable collaboration with veterinarian services, which, through their own technology, may deliver medication through an autonomous drone.
The Connected Car
Agriculture is just one of many industries using technology to more efficiently gather large volumes of data. Here’s another example, one that pertains to many adults the world over. Automobiles currently in production contain between 50 and 100 sensors that generate close to 5 GB of data per four-hour “shift.” These sensors collect everything from wiper blade speed to frequency of braking to patterns of acceleration. Unfortunately, most of this data gets discarded or becomes available only when the car is brought into the shop.
Until now, the primary use of a car’s connection to a data and communication infrastructure has been for safety and security uses. Such systems can detect when you have an accident, identify your current location, and automatically dial an emergency support line.
While all of the above serves a very useful purpose, the fact is that we miss the opportunity to exploit the richness of the sensor data, either on board or in an extended network. For instance, knowing how many cars are currently braking on a specific stretch of road is an immediate indication of road conditions. Such information could inform those currently approaching that area to beware. This data also could be a very accurate representation of the weather at that specific location.
The next wave of auto-based sensors will center on collision/accident avoidance. This wave will require the auto/driver combination not only to contextualize data being collected on board, but also to integrate information from vehicles nearby as well as the surrounding environment. This all leads, of course, to the driverless vehicle. It is estimated that the autonomous car will generate upwards of 700 MB of data per second, far exceeding the amount of data produced by current connected cars.
An autonomous drone flies to a predetermined location in a valley. The drone opens a small hatch and drops small, pill-sized, sensors along a road. The sensors activate, setting up a communication network that is able to detect and identify an oncoming convoy. The network also can determine the speed and direction of the convoy, communicating that information back to a base station. Is this a futuristic military capability? No. This test was successfully carried out in 2001 by The Defense Advanced Research Projects Agency (DARPA).
Since then, advances in technology have reduced these pill-sized sensors to the size of a grain of rice or even smaller. This model is known as Smart Dust and each individual sensor is referred to as a “mote.” Thousands of these networked and cooperating motes can be dispersed in many different ways: attaching to items (such as a seed), laying down in a pattern, or even casting to the wind.
The small size and low cost of these motes will make them completely ubiquitous and wildly useful for anything from physical security to measuring air quality or monitoring dangerous or hard-to-reach places. They will be programmed to operate independently, in groups, or even in large swarms. The amount of data streaming in from swarms of moving motes will be immense and beyond what we can imagine.
Going forward, such massive amounts of information will create significant challenges in understanding our increasingly quantified world. To exacerbate the problem, the amount of data being generated at machine speed will continue to grow at an ever-increasing rate. How do we harness this information and use it to its fullest potential? What do we use as contextual anchor points to exploit what amounts to billions of potential insights per second in a way that can be consumed by human or machine users? I believe Kant gave us the answer – in 1770.
Subscribe to Data Informed for the latest information and news on big data and analytics for the enterprise.