Threat Analysis and Risk Assessment, or TARA, is the analytical core of ISO/SAE 21434. Done well, it tells an engineering team which parts of a system deserve attention and why. Done poorly, it produces a large spreadsheet that is completed once, filed, and never consulted during design.
Scope discipline comes first
The most common cause of an unusable TARA is a poorly bounded item definition. If the item boundary is vague, the asset list becomes either exhaustive or arbitrary, and every subsequent step inherits that problem. Time spent agreeing what is inside the item, what interfaces cross the boundary, and what is explicitly out of scope pays back several times over.
Damage scenarios before threat scenarios
It is tempting to start with attacks, because attacks are concrete and interesting. The standard works in the other direction for good reason. A damage scenario describes the consequence for the road user: loss of vehicle control, unauthorised tracking, immobilisation of a fleet. Impact is rated against that consequence.
Starting from consequences keeps the analysis anchored to things the organisation actually cares about. Starting from attacks tends to produce a catalogue of technically interesting scenarios with no shared basis for comparing their significance.
Rating consistency matters more than rating precision
Attack feasibility rating attracts a lot of debate. In our experience the absolute value matters less than whether two engineers analysing similar scenarios arrive at similar ratings. Inconsistency destroys the comparability that makes prioritisation possible.
- Write down the rating criteria and worked examples before the analysis begins
- Calibrate with a small group on a handful of scenarios first
- Record the reasoning behind a rating, not only the value
- Review outliers as a team rather than adjusting them individually
Connecting the analysis to design
A TARA becomes useful at the moment its output changes an engineering decision. That means cybersecurity goals must be specific enough to derive requirements from, and risk treatment decisions must name an owner and a mechanism, not simply state that risk is reduced.
It also means the analysis has to be revisited. Architecture changes, features are added, interfaces move. A TARA that reflects the concept phase and is never updated will diverge from the product it is supposed to describe.
Written by
AutoSec Engineering Team
Automotive cybersecurity engineering
Engineers working on vehicle cybersecurity concepts, requirements, embedded implementation, and verification across OEM and supplier programmes.
More from this author