advanced Estimated learning time: 10 h

EC-20.1 Real Time as a Requirement: Deadlines, Latency and Jitter

You can state a timing requirement as a deadline with a consequence attached, measure the latency and jitter a system actually delivers, and say whether a superloop can meet the requirement or whether it needs a scheduler.

Before:EC-11. Data Communications and Computer NetworksEC-16. Ethics, Safety, Sustainability and Standards

Fast and on time are different properties, and most embedded trouble comes from confusing them. A system that usually responds in one millisecond and occasionally takes forty has a jitter problem, not a speed problem, and no amount of a faster processor fixes it. This topic makes timing a written requirement with a number and a consequence beside it, so that everything after it has something to be checked against.

Work through these

  • Hard, firm and soft deadlines, and what each one costs when it is missed

    A deadline is only meaningful when the cost of missing it is written down beside it. A missed frame in a display is a shrug, a missed sample in a motor controller is a burnt winding, and the same code is correct in one system and dangerous in the other.

    NPTEL: Real Time Operating System (IIT Kharagpur) · Course
  • Latency, response time and jitter, measured rather than estimated

    Latency is how long a response takes, and jitter is how much that time varies from one occasion to the next. Both are measured with a pin and an oscilloscope before they are argued about, because a worst case that has never been observed is a guess.

    NPTEL: Real-Time Systems (IIT Kharagpur) · Course
  • Periodic, aperiodic and sporadic work, and why the difference matters

    Work that arrives on a fixed period can be planned for exactly, work that arrives at unpredictable moments cannot, and work with a guaranteed minimum gap between arrivals sits between the two. Naming which kind you have decides which scheduling argument applies.

    NPTEL: Real Time Operating System (IIT Kharagpur) · Course
  • Worst-case execution time, and why the average is the wrong number

    A timing argument rests on the longest a piece of code can take, not the usual. Caches, interrupts and data-dependent loops all make the worst case much larger than the average, and a system sized on the average fails rarely and unpredictably.

  • Where a superloop stops working, and the symptoms it shows first

    A single loop calling everything in turn is adequate until one piece of work becomes long or one deadline becomes short. The first symptom is usually a response time that grows whenever an unrelated feature is added, which is the point at which the structure rather than the code is wrong.

    NPTEL: Real-Time Systems (IIT Kharagpur) · Course
  • Writing a timing requirement somebody else can test

    A requirement such as fast enough cannot be verified or refused. A requirement giving the event, the deadline in milliseconds, the fraction of occasions it must hold and the consequence of missing it can be tested by a person who was not in the room when it was written.

Sign in to keep your progress.

Free resources

Links last checked 4 Sept 2026.

Stuck here?

Ask a mentor. A real person answers, and they can see exactly which topic you're on. Usually within a couple of working days.

Checking your session…

Topics shown in module order.