Note: This article is the second in a series examining the lessons that have been learned through enterprise IoT projects in other industries and applying them to the connected car industry. To read the first article in the series, click here.
The rise of the Internet of Things (IoT) has caused tremendous growth in the number of connected devices and sensors in enterprises across all industries. In many ways, the connected car is the ultimate “thing” in the IoT. With hundreds of onboard computers continuously monitoring location, component performance, driving behavior, and more, connected cars are truly data centers on wheels.
As our transportation infrastructure continues to become even more connected through vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communication, connected car manufacturers and their partners will increasingly face many of the same IoT-related challenges that enterprises in other industries have already encountered and overcome.
With this series of articles (this is part 2), I’ll take a look at some of the best practices and lessons learned, and explore how they can be applied to the connected car market to help automakers and their partners harness the full potential of the IoT.
Lesson #3: Securing a larger number of attack surfaces
As connected cars continue to grow in popularity, another major challenge automakers face is security. Given all the data that connected cars can collect — everything from biometric and behavioral data on drivers and passengers to purchasing habits when the vehicles are used to pay for gasoline fill-ups and drive-through orders — there are implications for both privacy and safety if the vehicle is not properly secured.
On the enterprise side, as organizations have increasingly adopted IoT, they have wrestled with the potential new vulnerabilities of a greatly expanded attack surface.
Everything from an employee’s smartwatch to an organization’s IoT-enabled lights or security cameras can be turned into an attack vector. In the era of the IoT, enterprises have learned that there is no longer a security perimeter within which devices can be trusted. The IoT truly spans the globe and ecosystems of enterprises, even across industries. This requires a new approach to security, with defense distributed from the cloud to the edge of the network, as well as intelligence in the network to detect and stop threats before they propagate.
Not surprisingly, automakers are facing similar challenges. The attack surface in a connected vehicle is truly enormous — cellular, Wi-Fi, and even satellite connectivity between the car and the cloud, as well as V2V and V2I connectivity that cars use to communicate with each other and roadside infrastructure, can all be hacked. There’s also Bluetooth, near-field communications, and physical connections to the on-board diagnostics port under the dash. Even the wireless signals from the tire pressure monitoring system to the head unit can be hacked — the list goes on.
To secure this huge array of attack surfaces, automakers and their partners in the connected car ecosystem can adopt many of the principles and best practices that enterprises have used to strengthen IoT security. Converging and consolidating disparate networks in the vehicle onto a single architecture is one important step. With standardized in-vehicle networking, proven security technologies — such as encryption and authentication, firewalling, and intrusion detection and prevention systems (IDS/IPS) — can be deployed to give the connected car defense in depth. Artificial intelligence is also being used more frequently, both in the cloud and now in edge devices like connected vehicles, to detect new patterns of malicious behavior (or even non-malicious anomalies that could be early warnings of the need for maintenance).
Another key to securing all these potential attack surfaces is enabling the right levels of connectivity at the right times throughout the vehicle’s lifecycle. Much like how enterprise IT security teams have learned to continuously monitor their network access and IoT-enabled devices to spot potential trouble, automakers will need to continuously monitor and manage connectivity for their vehicles. They must know when connectivity for the vehicle should be on or off and what the vehicle should be allowed to do with that connectivity at different stages of its lifecycle.
For an automaker shipping millions of vehicles around the world, tracking and monitoring this connectivity is a complex task. For example, during the vehicle testing phase, connectivity must be “on” so that automakers can verify that connected services are properly functioning. Then, when the vehicle is in its shipping container, the manufacturer should automatically disable these services to prevent hackers from sabotaging the vehicle while it is en route to the dealership. However, some connectivity must remain on to enable real-time tracking of the vehicle during its journey. When the vehicle arrives at the dealership, an automated system allows automakers to safely resume connections so salespeople can demo the vehicle and its connected services to the buyer.
Lesson #4: Optimizing bandwidth, weight, energy, and other precious resources
As enterprises in other industries began adopting more IoT-enabled processes and placing an increasing number of sensors throughout their networks, they quickly realized that IoT presents several resource challenges. Not only are the devices themselves constrained (so that their applications need to be designed to make very efficient use of storage, compute, bandwidth, and power), but the sheer number of devices connecting can generate an avalanche of data, overwhelming networks and storage. Even before the IoT began to expand, proliferation of servers gave rise to a surge in virtualization, which greatly economized compute and storage resources in the enterprise.
Connected cars face many of the same challenges. Cost is an enormous factor for automakers looking to add new capabilities; energy is a precious commodity; and with regard to vehicle weight, every ounce counts. Moreover, highly automated vehicles can generate more than 4 terabytes of data per hour — vastly more data than a car can transmit over a cell tower network at any reasonable cost today. Nevertheless, highly automated vehicles demand more of these resources.
Here again, some enterprise IoT strategies apply. For example, while it would be nice to have all those terabytes of data from every vehicle available in the cloud as they are generated, some pieces of that data are much more relevant than others.
Moreover, some of the data is very time sensitive and some is not. Many connected vehicles have data plans where the data rate is cheaper at night, so determining when to send data is valuable. By applying intelligence to the in-vehicle network, automakers can determine which telematics data needs to be sent or which applications need to be connected at a given moment and which can wait until nighttime for a preferable data rate.
Likewise, the rise of fog computing (the practice of bringing the cloud to the edge with distributed compute and storage) in the enterprise enabled intelligent filtering and adaptive compression by IoT devices and gateways, thus reducing the amount of data that needed to be sent to the cloud. That same local processing in the vehicle, together with configurable rules about what data to send immediately versus what data to store and forward, can dramatically improve the effective use of mobile bandwidth in connected cars.
Finally, the “data center on wheels” can borrow another page from the enterprise data center notebook on the use of virtualization. Dozens of electronic control units throughout the vehicle can be streamlined by virtualizing some of their common logic, which will help reduce cost and complexity. Why have over a hundred small computers under the hood when you can consolidate many of them with an efficient, elastic, central computing platform that can also run the AI needed for highly automated vehicles? Virtualization and consolidation can also provide greater agility to create new applications and services from reusable components, reduce manufacturing and maintenance cost, and improve quality.
Lesson #5: Accelerate innovation through iterative rapid prototyping and a flexible architecture
Finally, but perhaps most importantly, the era of the IoT calls for rapid iteration and testing many ideas to see what works best and what delivers business value. Because of the radical reengineering IoT enables and because of the nearly limitless possibilities, rapid iteration and testing is especially important in the early stages of IoT, where results are often unpredictable. The automotive industry has been accustomed to a lengthier and more predictable cycle of innovation, planning major releases for new vehicles up to five years in advance and spending years testing every aspect. But because we’re still in the early days of both IoT and connected car adoption, it’s not always clear where a business will get the most value for their technology investments.
Automakers should therefore consider rapid prototyping to test new ideas quickly. They should also adopt a flexible architecture to shorten development and unit testing cycles, since the rigorous road testing cycles are more difficult to compress. Such architectures can make connected cars much more agile, allowing automakers to attach new sensors, actuators, or other devices; analyze the data; measure the value; and make adjustments very quickly. By reducing both the time to confirming valuable new features and the development and unit testing time, automakers can significantly shorten the innovation cycle to roll out industry-leading new releases.
The IoT is transforming enterprises in every industry, but the connected car is still in the early days. Fortunately, automakers can take many of the best enterprise IoT practices and apply them to connected cars to meet many similar challenges, including managing the complexity of a large number of connected cars each with myriad connected parts, while ensuring security and optimizing bandwidth, computing capacity, weight, and energy. This will help the automotive industry move to a future where vehicles will communicate with the cloud, each other, and the infrastructure around us to deliver a safer, more efficient, and altogether superior transportation experience.
Shaun Kirby is the director of automotive and connected car at Cisco, the multinational technology conglomerate.