The new buzz in aerospace is urban air mobility (UAM) – using urban low-altitude airspace for flying cars and drones. However, the number of aircraft, urban confinement, and aircraft design pose safety issues. An often-overlooked component in air safety is electronics bus technology.

A system bus can keep the aircraft in the air safe or bring everything down. It can affect electronics and software flow, triggering a fault tree.

UAM emphasizes electric and hybrid propulsion and total electronic control for autonomy and navigation. Many believe vertical takeoff and landing (VTOL) is the only solution, resulting in a complex system of propulsion and control. Integrating subcomponents of such a system demands reliable, predictable bus technology.

X-by-wire and sense-actuate loop design concepts are driving UAM. X-by-wire implies physical commands from the control pass reliably through interfaces to outputs such as thrust and control surfaces and provide feedback to a control pilot, autopilot, or ground command. A sense-actuate loop requires sensors, command inputs, and actuators seamlessly communicating with each other.

Though commercial aviation has long relied on Aeronautical Radio Inc. (ARINC) 429 and avionics full duplex switched Ethernet (AFDX) data-transfer standards, smaller aircraft demand lighter systems with less computing overhead.

Controller area network (CAN) buses have been used in aircraft and smaller drones for device communication. However, as the number of sensors and actuators increases, carrier sense multiple access (CSMA) is not enough to avoid data signal collisions. In UAM, responsiveness and real-time performance are critical. As communication nodes increase, responsiveness becomes unreliable, possibly leading to computational errors in the autopilot and other electronics.

The FlexRay network communications protocol’s fixed time-slots with time division multiple access (TDMA) technology offers an advantage for UAMs, but a new version is needed.


The ARINC 429 bus standard uses a twisted pair of wires with a grounded shield to transmit differential signals that are self-clocked but offer extremely low data rates. ARINC 429 isn’t a modern electronics bus with multiple transceivers communicating with each other on the same wires. A receiver pairs with one unique transmitter, although a transmitter can have multiple receivers.

Data is transmitted simplex (in only one direction). For bidirectional (full-duplex) communication, it needs two channels. The line replaceable units (LRUs) can be a transmitter, a receiver, or dual units. A transceiver in ARINC 429 is not bidirectional, but an LRU with each function side-by-side on an independent bus.


AFDX, patented by Airbus, is derived from the ARINC 664 standard with a profile of Institute of Electrical and Electronics Engineers (IEEE) specification 802.3 parts 1 and 2. AFDX offers the simplicity of an Ethernet bus-system and ability to use commercial off-the-shelf parts. Redundancy and network determinism with increased quality of service was added.

AFDX offers a packet-switched and profiled network, full-duplex communications, and a higher data-rate. It enables simpler cabling by using virtual links (VLs) – logical notions that connect a pair or multiple devices in a simplex sense such as unicast or multicast data packets. Switches at the end-systems perform traffic shaping, traffic policing, and static frame routing. Full-duplex communication is achieved by two VLs between the device pair.

Least-examined with AFDX are the methods of bringing determinism in the network. End systems manage traffic shaping and VL integrity checking, while switches perform traffic policing. The design and nomenclature of the policy-shaping and policing for deterministic behavior prevent analytical verification.


In CAN, a primarily CSMA-based bus system, data packets are rescheduled with every collision sensed. CAN assumes bus arbitration solves conflicts and achieves real-time performance. Node priority is assigned mostly by trial and error, so symptoms get more pronounced when the number of nodes and frequency of information exchanges increase.

Cyber-physical systems are not computer networks, so while CSMA algorithms could diminish dead-time, they could also diverge the time forcing each node to wait for a non-deterministically long time. This affects safety-critical functions and can trigger a bad fault tree. CAN variants – such as CAN HS at 1Mbps, CAN FD at 10Mbps, and FT at 125Kbps – trade performance for data rate, yet the underlying CSMA can’t offer complete determinism.

The control-loop bandwidths needed for urban, space-constrained flight, preclude CAN as a candidate for responsive, safety-critical systems.


The FlexRay consortium’s architecture could be transformed to a bus standard for UAM. FlexRay includes physical layer circuit redundancy and offers 2-wire or 4-wire connections, reliability, and determinism. It is also topology agnostic.

FlexRay achieves repeatable determinism with time-triggered TDMA and universal clock reference. The behavior cycles of every component and subsystem can be quantified and the central computer can assess time cycles of each component and subsystem to inform functions and methods of overall flight-control software. Time-triggered TDMA assures repeatable, real-time performance.

Modifying FlexRay for UAM

To build a network of subsystems and components, consider:

  1. FlexRay uses low-voltage differential signaling (LVDS) for noise immunity, however, electric noise that could be experienced by UAM airborne systems, without a physical ground, must be assessed. Increasing the voltage levels could decrease noise susceptibility, but voltage level and data rate are a tradeoff in a bus with intersymbol interference. Direct current buildup in the cables can spoil the margins of transmitter and receiver. Agreeable voltage levels and corresponding bit-timing must be developed.
  2. Weight of cables and connectors must be repurposed for drones and flying cars.
  3. Redundant channels should be strictly maintained only for safety, not for increasing the data rate.
  4. The baud rate should be fixed for drones at 2.5Mbps and 5Mbps for flying cars.
  5. While FlexRay can support multiple topologies and the bus can support up to 64 nodes, there should be a limit of 16 active star networks in the vehicle to maintain accuracy within a few nanoseconds for wakeup, startup, and synchronization procedures.


A derivative bus technology shouldn’t be a significant departure from the existing standard allowing cost-effective verification and validation. UAM will likely be a regulated industry, and the software and hardware need to comply with RTCA/DO-178B (or EUROCAE ED-12B) and RTCA/DO254 respectively. DO-178 is a formal software certification and DO254 asserts a process on hardware such as programmable logic devices (PLDs), field programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), and programmable analog arrays (PAAs).

A state-transition model makes it difficult to predict transition time performance, but with the right timing diagram, system transitions can be observed with a fixed time quantum. FlexRay-based design inspires such a design. It forces us to incorporate timing metrics in hardware and software, producing a system with inherent safety and possibly easier certification.


About the author: Seshu Kiran GS, founder and CEO of X Air Sky, can be reached at 310.237.5562 or