As the convergence between vehicles and computers continues, with the end result being software-defined vehicles (SDVs) becoming more common on the road, exciting new capabilities such as real-time over-the-air (OTA) updates, advanced driver assistance systems, autonomy, detailed infotainment, and more are enabled.
However, accompanying that convergence is a currently unsustainable rise in bandwidth demand. The looming SDV era will see these surging data requirements that present both a challenge and a unique opportunity for the automotive industry. Consider solely the function of autonomy: “full” autonomy (that is, when all of a vehicle’s passengers can safely divert their attention from the road in transit) will require upwards of 200 Gbps.
One proposed solution is automotive Ethernet, which, if broadly adopted, holds the potential to keep vehicular data demand in pace with processing capabilities—one day enabling a fully realised autonomous vehicle.
The five levels of vehicular autonomy
When most hear the phrase “autonomous vehicle,” they likely imagine a passenger car fully capable of navigating itself while the ‘driver’ is otherwise predisposed or entertained. However, that type of autonomy is maximalist (and furthest away from our current reality). Achieving that level of autonomy requires incrementalism, which is why both passenger and commercial vehicular autonomy alike exists in five levels:
- Level One, Basic Driver Assistance: The most common level of ‘autonomy’ on the road today, basic driver assistance includes features such as adaptive cruise control, emergency braking, lane departure warning, blind spot warning, and parking assistance. Level one requires upwards of 2 Gbps total bandwidth.
- Level Two, Partial Automation: At this level of automation, the automobile adds expanded parking assistance as well as lane “keep” assistance to supplement the lane departure warning. This level can be considered ‘feet off,’ and requires upwards of 32 Gbps.
- Level Three, Conditional Automation: Level three automation is the ‘hands off’ level, adding sign recognition, collision avoidance, cross-traffic alerts, surround view, and rear-collision warning. This level sees a jump in bandwidth demand to 48 Gbps.
- Level Four, High Automation: Level four enables ‘eyes off’ the road for all passengers in a vehicle, adding speed-limited autonomy. Achieving such a high level of automation is expected to require up to 95 Gbps of bandwidth.
- Level Five, Full Automation: Level five allows all passengers to fully take their ‘minds off’ the vehicle’s actions entirely as a fully autonomous system completes transportation. This level, as mentioned above, will require the highest bandwidth—up to 200 Gbps.
The limitations of automotive architecture today
The five levels of autonomy provide a neat roadmap of technological innovation towards the end goal of safe, reliable, efficient full autonomy. However, domain architecture, the average vehicle architecture on the road today, is not equipped to support the data demands of future autonomy levels.
A proposed solution
Automotive original equipment manufacturers (OEM) must instead create an architecture that supports the trend towards great autonomy today. One solution is a centralised architecture with a network that can support access to all of a vehicle’s data.
Such an architecture includes zonal aggregators that gather data from different domains into centralised compute ECUs. That architecture follows the same logic as domain architecture but significantly reduces complexity, weight, and cost by using Ethernet as the interconnect. It also unlocks greater levels of security, reliability, functional safety, and, of course, autonomy.
The trend towards autonomous, SDVs continues to build momentum. A report from Global Market Insights valued the SDV market at $35.8 billion in 2022 and projected a compound annual growth rate (CAGR) of 22.1% between 2022 and 2032.
McKinsey, meanwhile, recently projected autonomous vehicles “could generate between $300 billion and $400 billion in the passenger car market by 2035.”
While exactly when and how advanced features like autonomy will be developed and implemented may be debated, it is clear what both consumers and the industry expect in vehicles a decade or two from now; but, the car of the future will never be built if car architecture does not sufficiently evolve first to meet bandwidth demand. Solutions like the centralised architecture described above must be sufficiently advanced soon, or else the industry risks delaying the car of the future—or worse: producing a future car that is unsafe and unreliable.
Author: Ramin Shirani is CEO and Co-Founder of Ethernovia.
