Autonomous vehicles have captured the imagination of futurists, manufacturers, planners and the public. Driverless vehicles promise new levels of road safety and convenience – who would not prefer reading the newspaper to micromanaging stop and go commuter traffic – while offering potential for better management of traffic flow, an increasingly urgent problem in many cities around the world. Anticipating this new car vision, virtually all major auto makers now have an autonomous vehicle program in place, as do several of major tech firms, including Apple and Google. Market researchers at CB Insights recently identified over 40 global companies that are investing in autonomous vehicle R&D, and though manufacturers pose different time frames for the advent of driverless cars, many predict that some level of self driving will be available by 2021, and ‘fully autonomous’ vehicles will be ready for the road in urban settings by the end of the decade.
The enthusiasm surrounding the introduction of autonomous vehicles (AV) assumes the resolution of several key issues, ranging from safety – the fatal accident this year caused a pause in program roll out for Uber – to the establishment of liability in the case of accident and the insurance industry’s creation of new policy approaches. But it also assumes that the infrastructure needed to support autonomous vehicles will be in place, a proposition that is less than certain. Like the electric vehicle revolution, which faltered on driving range, i.e. the ready availability of charging infrastructure, the evolution of autonomous vehicles is contingent on access to the seamless connectivity needed to power Internet-based systems that will support the connected car. Connectivity is a requirement at virtually all levels of autonomous vehicle operation; while simple driver assistance systems leverage cell and radar services, it has been estimated that a Level 5 car will send 25 gigabytes of data to the cloud every hour.
In Canada, this challenge is being tackled on a number of fronts, and with good cause. While mobile penetration rates in the country are relatively high overall, provider business models demand the delivery of most coverage in population dense urban regions. Connectivity in Canada is clustered along the country’s southern border, leaving many more remote and northern regions underserviced; while service has been described as “bursty,” or lacking in the two way symmetry needed for interactivity. Dropped cell calls and lack of broadband service are an experience that many Canadians have shared, and a daily occurrence for residents in more sparsely populated regions. Autonomous vehicles, however, require reliable, quality connectivity that can power the latency sensitive applications and systems needed to maintain vehicle safety. To support the development of appropriate levels of connectivity, several government-sponsored initiatives are underway. For example, with funding from the Ontario government, the Autonomous Vehicle Innovation Network (AVIN) program is conducting research into autonomous technologies, including networking, while providing a connected testbed for local companies to validate commercial AV solutions. Similarly, the Ottawa L5 test facilities launched in 2019 are equipped with GPS (RTK), dedicated short range communications (DSRC), WiFi, 4G/LTE and 5G telecommunications and networking infrastructure to provide an integrated, connected autonomous vehicle test environment for companies to prove out connected and autonomous vehicle (CAV) solutions.
Designed to foster innovation in advanced CAV applications, these facilities are beginning to show results. A good example is found in a proof of concept (PoC) developed by Martello Technologies in collaboration with BlackBerry that was showcased on the L5 Track this past May. Presented to the Canadian Association of Chiefs of Police, the Martello mobile network performance PoC demonstrated that it is possible to maintain reliable cellular network connectivity for bandwidth-intensive real-time services, as would be required by autonomous vehicles. In addition to AV environments, the demo highlighted Martello technology capability to provide solutions for mobile office, autonomous trains and drones, or disaster relief and emergency response – applications where reliable communications are critical to safety.
In the video presentation shown below, Martello connectivity is marshalled for the car’s infotainment system, but as president and CEO of Martello John Proctor explained, the solution supports ‘connected car’ more broadly, opening the door to new business models in the automotive space. Proctor believes we are years away from true autonomous driving, due to issues with governance, certifications and the AV’s current inability to make ethical decisions. But there are multiple new opportunities in the automotive industry today, that are now enabled by advanced connectivity: “if you look at it from a vehicle manufacturer’s perspective, they already know there’s enormous value in being able to connect the system, and retrieving data from the car and using it to find another revenue stream. If we're going into this sharing vehicle economy, they [manufacturer] may well become the insurance company because they can assure you based on your driving behavior: the car will know. So why would they give that data away to an insurance company? They're going to become the insurance company themselves.” Proctor cited other use cases – restaurant recommendations, for example, that the car offers as the passenger travels in the vehicle, based on predictive analysis of the driver/passenger behaviours. “All of that requires connectivity,” he added, “and we haven't even touched on the vehicle itself, which will want to have traffic updates, smart city updates, or map updates. It's not just the entertainment system, it's connected vehicle and we are the “connect” in that connected vehicle aspect.”
The Martello PoC also demonstrates the potential that may be realized through technology partnerships. Developed over a six month period in collaboration with BlackBerry QNX, as part of the BlackBerry QNX/L-Spark Accelerator program, Martello’s solution entails embed of a virtualized version of the technology inside the QNX operating system. Martello’s interest in working with the Blackberry platform was not limited to Blackberry leadership in the AV OS market. According to Proctor, the Spark program also provided access to senior tech folks, for discussion of real and current challenges. “We weren’t building something in hope they would come,” he explained. “We were able to say to them, ‘what is the business problem you currently have, what challenges do you see in the future, and can you articulate the issues clearly to us because we think we can solve them.” By building to address identifiable problems, Martello positioned to approach QNX partners, such as the auto manufacturers, with a workable solution, while providing Blackberry with value added capability that could be offered to upsell the QNX operating system.
The secret sauce in Martello’s “connect” are Service-Aware networks that rely on performance analytics and testing to identify faults, and on SD-WAN to optimize performance of network resources in real time. In an AV scenario, this translates to functionality designed to ensure multi-link redundancy – or the use of sophisticated algorithms that can leverage multiple network provider services to ensure a consistent connection. As the company website notes: “two-second delay in decision-making for an autonomous vehicle can be catastrophic.” Proctor explained: “we wrote algorithms that take all the noise out and enable connection to three cell towers simultaneously, which provides redundancy. So as number three signal fades, we pick up number four and as number one fades, we pick up number five. The way Canada is covered, you cannot do that with just one provider. They [providers] have literally put their cell towers out, as they should, to try and cover the maximum ground. It’s not always perfect. But if you can do cell tower hoping while maintaining coverage, it makes a significant difference.” Through this bonding of provider services into one Internet connection, Martello not only helps to deliver good coverage, but also enables traffic optimization. Martello technology can identify the best path for the type of information that will be sent. “There’s quite a difference between real time connectivity or connections for taking a call in the car, for video conferencing, or when the car needs real time updates from a smart city, and or you are going to send emails,” he noted. “That forces the need for multiple redundancies.”
Additionally, in northern or remote regions, communications systems will need to connect via different networks – perhaps through a satellite receiver on the car roof as there is no cell coverage that connects into the Internet in these regions. And while 5G is now touted as the platform that can resolve these kinds of issues, Proctor advises that it is unlikely that Canadian regions outside major urban areas will see 5G coverage for a long time. In the meantime, the car needs to connect, and in so doing, needs to be able to access all available means of communication, different links and Internet connection paths, and to optimize the traffic that crosses them. The Martello solution works with all available networks, including several generations of cell, satellite, DSL, microwave, combining these into a larger connection that can approach 30 Mbps.
According to Proctor, a key to Martello success is connection at layer two in the network: “our competition connects at the layer three, where you have to figure out and finalize the IP address, you have to reconfigure the router, and you have to redo the firewall rules.” Martello’s connection at the data layer, he argued is much easier. In addition, it’s more secure as it’s not possible to feed off the connection, and it's easier to install. Layer three connection is “kind of funky,” he observed, and requires the help of a smart engineer for installation.
Proctor sees new opportunity for the car manufacturer in roll out of continuous connectivity. While Martello has created an appliance that will accept multiple SIM cards from the different cell providers (Bell, Telus and Rogers), which could be deployed as a remote terminal on the car to service end user connectivity needs, Proctor describes this as a “clunky, ad hoc” solution. A better approach in his view, is for the car manufacturer to deliver this capability as part of a new, software SIM-based service model: “In theory,” he added, “they could get to the stage where they say to end users, ‘oh I can see you’re taking this eight hour journey, because you’ve just put that you're taking this route into the map. We work with Martello and know that you won't have coverage for this section – if you buy four hours of this software SIM card, you can get coverage from Videotron, or from potentially from a different provider.”
This kind of ‘connectivity-as-a-service’ promises to simplify the journey for end users, would be an ultimate requirement in AV. But it demands new monitoring and pricing capabilities that can track usage, whose connection is being tapped, and also, who pays for service. Adopting a cloud-like, usage-based model, Martello has managed this challenge in other client segments, and believes the approach could be adapted to connected car. The hotel industry, for example, has shifted from the sale of movies on TV to the sale of an Internet connection in standard or premium service formats, and depending on whether a guest wishes to do simple email or stream video, there are corresponding charges. “It is our devices that recognize if you pull the premium or not, and then give you access to whichever amount of bandwidth you have,” Proctor explained. “We already do that, and so it’s a fairly simple transition to bring something we do on a daily basis with thousands of hotels to connected vehicles.”
Martello has seen considerable success in the hospitality space – it counts clients ranging from the Fairmount Royal York in Toronto to hotels in Afghanistan – due to capability and cost. According to Proctor, a Martello appliance is much cheaper to install and operate than a comparable device from the company’s US-based competition because installers do not need engineering certification, and because as a smaller entity, Martello runs in a lean fashion without a lot of overhead. Other verticals where the company is beginning to see traction include the retail industry, which is cost conscious and may lack adequate IT resources for complex installs, and financial institutions outside Canada, where Martello has tested alongside the large network equipment providers and prevailed.
Today, connected car is a future application opportunity for the company. But by refining its network performance management capabilities in other verticals, Martello continues to position for the larger market potential that IoT applications will deliver. Martello’s story is building. Following its IPO in the fall of 2018, the company has experienced keen interest from traders: Proctor claims that Martello was the most heavily traded stock in Canada for one week in July. Through channel partnerships with established firms like Mitel (for its UC offering), and new relationships and acquisitions, the company expects to continue a growth trajectory which saw Martello report a revenue increase of 103% for fiscal 2019 over the previous year. Proctor attributes much of this success to activities in global regions and looks forward to closer attention in domestic markets. Financial results “really speak to the fact that people have had a chance to look at Martello, recognize that we are a very viable Canadian tech story,” he noted. “We’re kind of the little Canadian tech company that can. We can achieve large wins with other governments, with other giant enterprises, but tend to get ignored in our own backyard.” Based on Martello capabilities and growth, it’s likely this will be a shorter story.
 IoValue: Intelligence in Community Ecosystems. An IoTCC Best Practice Report. March 2018. pp. 15-16.