Embedded System Design Cycle Questions and Answers

Answer: A

Every later decision, the part included, follows from the requirement, and a part chosen first constrains the requirement rather than serving it. Requirements written down are also what a test can later be traced back to.

Answer: B

Response time, power consumption, temperature range and mean time between failures are all of this kind, and they constrain the architecture at least as much as the functional list. They are also the requirements most often left unstated and discovered late.

Answer: C

The matrix answers two questions: is every requirement implemented and tested, and does every piece of the design exist for a reason. Certification bodies ask for exactly that evidence, and it is far cheaper to maintain than to reconstruct.

Answer: D

Correct answer: 2, 4, 3, 1

Requirements come first, because nothing can be designed until what the product must do is settled. Architecture and detailed design follow, then implementation with unit testing of the parts, and finally integration and system test of the whole. The order is forced by dependency, and it is the same whether the process is run once as a waterfall or many times in an iterative cycle.

Answer: A

Each level of decomposition on the way down has a matching level of verification on the way up: unit tests against the detailed design, integration against the architecture, acceptance against the requirement. The shape makes it plain what each test is testing against.

Answer: B

A board respin takes weeks and costs money, so the hardware cannot iterate at the software's pace. The usual answer is to iterate the software freely on a development board while the hardware is committed in a small number of deliberate steps.

Answer: C

Correct answer: which functions are realised in logic and which in code

Partitioning is the decision about where each function lives - in a processor executing instructions, or in dedicated logic in an FPGA or ASIC. It is taken early because it determines the architecture, the cost and much of the schedule, and it is one of the few decisions in an embedded design that is genuinely expensive to revisit later.

Answer: D

Correct answer: must meet a timing requirement the processor cannot

Hardware is chosen when the timing cannot be met in software - a response measured in nanoseconds, a sustained data rate, or many operations that must happen simultaneously rather than in sequence. The other considerations point the other way: a function likely to change belongs in software precisely because software can be changed after the product ships.

Answer: A

Updating firmware over a communication interface is routine; reconfiguring fabric in the field is possible but far less common and carries more risk. Timing and power both favour the fabric, which is the trade being made.

Answer: B

The algorithm is developed and tuned against a model of the physics, which is far quicker and safer than a real machine. Software and processor in the loop then check the generated code, and hardware in the loop puts the real controller against the simulated plant.