Konstantin Sutyagin - Fotolia
When people think of IoT, their minds typically drift into thinks like the interconnected home, smart drones and autonomous devices. They don't typically think about transport trucks. But a short discussion with Brad Taylor, VP of Data and IoT Solutions at Omnitracks, just might alter your perceptions. His organization is leveraging big data IoT projects to serve a good number of trucking industry clients, all in preparation for an increasingly data-driven future.
"The average truck driver drives 125,000 miles per year. Over 90% of goods that arrive in a grocery store or at your house spent some time on a truck," said Taylor. "These are assets that are on the road all the time and they are essentially big gateways at this point. A lot of people think of a Tesla as being the most automated vehicle. But in fact, we were connecting trucks over two decades ago." That's a lot of trucks delivering a lot of big data.
Exploring the four Vs of big data—plus one
As with all data, the four Vs of volume, variety, velocity, and veracity are important when it comes to making sense of the connected transportation industry. With Omnitracs handling over 500 million transactions per day, big data IoT volume is obviously high and will only increase. Velocity is also significant. Trucks are always on the move and delivery routes and schedules are constantly changing as well, so the organization requires both real-time and micro-batch processing to make sense of the data. Variety is substantial with data arriving in SMS, MMS, GPS, video and other formats. Because the company is dealing with so many data sources, veracity in terms of consistency and quality of data is vital as well.
"These are some of the things we were looking at from the "four V" perspective, but what about the Value perspective? How do we make decisions that help us move forward with our business?" Both relevancy and uniqueness played a role in determining the value of IoT big data. "Timing is important for us. If the data is late, it is no longer relevant. Location is important, because a minute later, you're a mile away. The data can go bad very quickly."
Brad explained that data has value to Omnitracs both in a unitary and aggregate sense. "One data point offers very precise information. It's a variable that tells you something absolutely unique. For example, if you know the ignition is out on a specific truck, you know that vehicle is not going to start." On the larger scale, the more data one has, the greater the value. "The Russians have a saying: Quantity is its own quality. If you have enough data, you'll find good data to use effectively. A large volume of data can also be useful to someone outside your organization if there's enough of it and it matches what they are looking for."
Smart data increases the IQ of big data IoT projects
As Brad pointed out early in his presentation, "Big Data isn't just big anymore. It may be fast data and it may also be smart data. So, when you are looking at the value of your data and trying to determine a strategy, it's important to understand what you really have. It can impact how you design your architecture." Omnitracs considered use cases for each of these three characteristics in making architectural and business decisions.
Making data smarter started with taking a look at the most fragile component in the transportation equation: the driver. With too few drivers for the loads that needed to be shipped and shockingly high turnover of up to 75% per year, it was an area ripe for analysis to prevent unwanted outcomes. Previously, Omnitracs gathered data from customers and started from that point in determining which predictors to use. However, searching through these 'stacks of hay' and building custom prediction models with disparate data sets was an expensive process.
Brad's people needed to be looking at better quality or smarter data to deliver valuable results as quickly as possible. The answer was to obtain a smaller number of higher quality data sets. Omnitracs focused on data from highly regulated sources and those under their own control instead of relying on data from customer sites. This approach actually required no architectural changes. As a result of this decision, they now have smarter data that highlights at risk drivers who are most likely to be involved in preventable incidents.
Big data IoT projects promise speedy ROI
In an industry like transportation, time is of the essence. That's as true in the digital world as it is on the road. While some businesses are eager to build a data lake, Omnitracs needed to manage a data river, one that was never the same from moment to moment. "A data lake is data at rest. Data in motion is much more important." According to Brad, the speed vector data they collect accounted for over 100 billion data points a year and that was just one of the many types of data being garnered. This incoming data needed to be analyzed within two minutes in order to be relevant.
That's because decisions about where to drive may have to happen very quickly in changing road conditions. "With trucks, you need to take into account the parameters such height and weight restrictions as well as regulatory restrictions on where they are allowed to drive. You need to understand these things so you don't lead the truck into a location where there is a hazard or a fine." Being able to analyze big data IoT project information immediately to ensure routing and navigation was critical. Tools like Flume and Kafka played a role in making the company's data processing architecture faster.
Driving big data is about to get much bigger
The price for an accident is steep in the trucking industry, and Omnitracs' clients are highly motivated to find ways to avoid those high payouts. Understanding how accidents happen is a vital part of the process. And the data that's available right now is just the tip of the ice berg. "As we started to aggregate the data, we found that a lot of our predictive models had been based on the assumption that things like hard braking and over-speeding were representative of what was important."
When they began looking at the data in detail, they found that ninety percent of critical events were associated with the fifteen percent of trucks that featured advanced safety systems such as lane keeping and distance following. These events were tangible, but not material. In other words, all trips might include a variety of suboptimal driving behaviors or near misses. But it was only the advanced safety and reporting features that made these facts visible for analysis. Soon, visibility is set to reach unprecedented heights. "In the coming years, all new trucks will be required to have certain safety features. Omnitracs is preparing for a deluge of new data."
The future may also include inward facing video cameras to complement those cameras that face outward to record what's going on in the driving environment. Taylor suggested that it will be interesting to see the Hawthorne effect in action, revealing what changes when drivers know they are being observed.
Of course, all these new data inputs will require even bigger, faster, and smarter solutions. Like many organizations that have data driven revenue, Omnitracs sees this as simply another opportunity to deliver greater value to a larger set of customers more quickly than ever before.
Learn more about big data IoT projects and data warehousing
Here's Gartner's look at the hype around big data IoT projects
Find out why big data and analytics need to share the stage