The automotive industry is witnessing a transformation from traditional mechanical and electronics component-heavy architecture to a software-driven paradigm commonly referred to as Software-Defined Vehicles (SDV). A similar transition occurred in the telecommunications sector about 15 years ago when service providers and OEMs moved from hardware-centric architecture to a flexible software-driven ecosystem. It is only natural to explore the SDN technology transformation and apply its learnings to the SDV world. Let us first understand the parallels between SDN and SDV through a multi-dimensional lens – Architecture, Network, Compute, Security and Standardization. These five perspectives create a framework that allows us to compare and contrast the two worlds and abstract valuable learnings to avoid making the same mistakes.

Software Defined Architecture: The networking world saw the decoupling of control and data planes, and moving the control logic to a central controller.

• Further, Network Function Virtualization (NFV) migrated traditional hardware functions to software.
• In SDN, abstraction allows standard commands to configure the underlying hardware. In SDV, SOAFEE, and COVESA achieve that through VSS (Vehicle Signal Specification).
• Lastly, there is a migration from Signal-Oriented architecture to Service-Oriented Architecture based on a Publish/Subscribe mechanism.

Network Transformation: Traditional in-vehicle automotive networks are low-bandwidth, shared medium and are not designed to handle SDV load.

• Automotive Ethernet provides for high bandwidth, switched fabric topology with the zonal controllers acting as gateways. The connectivity to actuators/sensors still uses CAN or LIN.
• Dynamic reconfiguration of network load allows for prioritization and traffic routing based on the vehicle mode, parked or autonomous, for instance.
• Encapsulating data packets from legacy sensors or actuators through the CAN or LIN protocol and decapsulating in the HPC, a technique borrowed from SDN.

Centralized Compute: The migration to a central high-performance compute supported by zonal controllers is similar to the SDN software logic.

• The zonal architecture with controllers at the zone serving as switch/gateway and processing by the HPC reduces wiring weight and increases updateability.
• In the SDN world, Virtual Network Function (VNF) helps operators test new configurations before rolling them into a new network. vECU enables similar shift-left testing.

Security and Safety: One key difference is that telecom networks are designed for high availability, whereas automotive networks are all about safety.

• Dynamic reconfiguration of SDN to ensure availability will be constrained in an SDV world. In case the system enters an inoperative state, it must retreat into safe mode.
• SDV increases the attack surface, and hence there is a need for additional security mechanisms – distributed firewalls, message authentication, and Intrusion Detection Systems.

Standardization and Ecosystems: SDN adoption was driven by open standards, whereas the automotive industry is fragmented when it comes to standardization.

• Adaptive AUTOSAR has been designed for high-performance compute and Service-Oriented Architecture; however, the criticism is that it is very slow to evolve.
• SOAFEE is an industry-led initiative to bring cloud-native development standards to automotive engineering, and there is a conflict with the safety-first culture of AUTOSAR.
• COVESA standardizes the data specifications, and specifically the Vehicle Signals, so that the abstraction can be used across vehicles.

As we can clearly see, SDV embraces decoupling of hardware and software, virtualization to minimize hardware dependency and SOA to achieve modularity from the SDN world. However, there are differences in terms of telecom networks’ focus on availability, and best-effort basis, whereas automotive is deterministic and safety-first. Here are some key learnings from SDN implementation that the automotive ecosystem can benefit from. In fact, the automotive industry must embrace practices from successful implementations across industries.

• Differentiate through software, hardware commoditization is happening across industries
• Standardization is critical, and the industry must coalesce around open standards
• Security is not an afterthought, but is structurally built throughout the entire process
• Foster ecosystem collaboration with suppliers and technology partners
• Manage mixed-criticality workloads while efficiently sharing resources

Software-Defined Vehicles is a Venture under the Infosys Business Incubator capability. The Infosys Business Incubator (IBI) is a strategic capability aimed at fostering entrepreneurship within Infosys. It empowers employees to ideate, incubate and create new businesses. Know more here.