Issue 39 Unmanned Systems Technology August/September 2021 Maritime Robotics Mariner l Simulation tools focus l MRS MR-10 and MR-20 l UAVs insight l HFE International GenPod l Exotec Skypod l Autopilots focus l Aquaai Mazu

90 share their transponder data – regardless of what type it is – for anyone to access and receive, be it an autopilot or UAS operator with a smartphone. The system works over the internet without the need for software or licence- based restrictions, and comes with three levels of access – a basic, public level where a UAV’s position and heading can be viewed; a more secure and privileged level of access through which operators and potentially UTM authorities can intervene; and a topmost layer of high-integrity real- time data for landing authorities to access for managing incoming aircraft. Cyber security The dangers of populating cities and airspaces with unsecured autonomous vehicles have been discussed in security conferences for well over the past decade. One expert says the current state of cyber security is such that hackers might find it somewhat easy to remotely seize a consumer or commercial UAV for nefarious purposes. By and large, incorporating AES-256 encryption in data links has been the extent of cyber protection in unmanned systems. Failure to go any further than that leaves open an array of ingress points that hostile actors could use to conduct espionage or seize the vehicles, as is the case with any computer system. Defence departments nowadays are increasingly calling for improved cyber security measures regarding how UAV components are designed and engineered, with higher levels of autopilot protection at the heart of these to address the severe gap in how unmanned systems can be kept out of the hands of criminal or terrorist groups acting from afar. This can mean designing an autopilot from the ground up with the necessary components and architecture for cyber security. One of the most prominent approaches to this involves building the autopilot around a very powerful processor, one that is significantly over- specified compared with the minimum hardware requirements needed for the core functions of flight control. Not only does that remove the need for tactical edge computers to be interfaced alongside the main autopilot for carrying out AI- or cloud-related tasks – which can pose serious vulnerabilities to the rest of a vehicle’s onboard systems – it also leaves more than enough processing capacity for running cyber security functions. These functions consist largely of encryption. That means not simply encrypting packets of data at the vehicle’s radio before it is transmitted through the air, but encrypting every piece of data before it comes or goes from the autopilot or the GCS, to a standard such as AES-256. Encrypting every packet of data running throughout a UAV’s network bus helps enormously in assuring that nothing has been altered by some form of cyber attack. Additional functions include applying and checking digital signatures on data to assure the integrity of its contents and timing, authenticating users who want to access autopilot information, and verifying the security of boot processes. Having a processor powerful enough to do all that alongside flight control, guidance and so on runs counter to the less expensive architectures of open-source systems but it can prevent key points of vulnerability arising while leaving space for configurability that proprietary autopilots might fall short of. Some guiding standards are available for informing how an autopilot can be designed to meet governmental and defence expectations on cyber security. For example, FIPS (Federal Information Processing Standards) 140- 3 stands out in the US as a tangible set of requirements a piece of technology must meet to be able to handle classified information. FIPS 140-3 lays out four levels of increasing security designed for assurance over cyber security across a broad variety of possible use-cases and operating environments, in order to inform the approach to cryptography when designing and implementing secure software modules. Its text includes details of specifications, interfaces, authentication, firmware security, physical security, self-tests, life-cycle assurance and many other important facets of cryptographic modules. That gives software engineers a clear set of protocols and levels they can adhere to when writing security software for an autopilot. FIPS 140-3 is August/September 2021 | Unmanned Systems Technology The cyber security of unmanned systems is becoming a point of concern; standards such as FIPS 140-3 provide critically useful guidelines for how software engineers can program an autopilot for high cryptographic robustness (Courtesy of Asymmetric Technologies)