EE8405 Chap.3 Poisson Processes and Exponential Times
Poisson Processes and Exponential Times
A Poisson process describes an event stream through counts in intervals and exponential times between events. The chapter translates between those views, aligns rate and time units, combines independent streams by superposition and analyses competing exponential clocks.
It also treats stationary and independent increments as assumptions that must be checked against burstiness, trends and serial dependence in real telecommunications traffic.
What this chapter covers
- 01
Poisson counts
- 02
Exponential waits
- 03
Memorylessness
- 04
Superposition
- 05
Thinning
- 06
Competing clocks
Worked example · free
Link a count to a waiting time
- 2Convert the interval into the rate unit.
- 2Evaluate the Poisson event.
- 2Check the zero-count case with the exponential tail.
Key terms
- Stationary Increments
- The count distribution depends on interval length rather than its location.
- Memorylessness
- Elapsed waiting does not change the exponential distribution of remaining waiting time.
Poisson Processes and Exponential Times FAQ
When is a Poisson model plausible?
It is most defensible for independent events from many sources at a roughly constant rate. Trends, coordinated retransmissions, batches and long-range dependence can violate its assumptions.
How do independent Poisson streams combine?
Their superposition is Poisson with a rate equal to the sum of component rates. Independent thinning produces subprocesses with rates scaled by their classification probabilities.
Why must lambda times time be unitless?
The Poisson parameter is an expected count, so rate and interval must use compatible time units. A dimensional exponent signals that the conversion has not been completed.
How can the count and waiting-time views check each other?
The event that the next arrival occurs after time t is the same event as observing zero arrivals during the next interval of length t. Evaluate the Poisson zero-count probability and the exponential survival probability with the same rate and time units; they should agree.
This check catches a wrong rate parameter, an hours-to-minutes error and a confusion between density and survival probability before later queueing formulas inherit the mistake.
What evidence would make a Poisson assumption doubtful?
Trends violate stationary increments, correlation across disjoint intervals challenges independent increments, and substantial batches contradict the small-interval single-event behaviour. Compare mean and variance of equal-window counts, inspect lag correlation and plot rates across time. The diagnosis should identify a mechanism such as sessions, retransmissions or scheduled demand.
A mismatch calls for sensitivity analysis, simulation or a richer arrival model rather than an unsupported claim that every Poisson result is useless.
What is the fastest unit check for a Poisson calculation?
Rate multiplied by interval must be dimensionless and represent an expected count. Convert time units before evaluating either the count probability or exponential survival. Label the interval explicitly.
Exam move
Solve zero-arrival events twice, once as a Poisson count and once as an exponential wait. Use agreement as a unit and parameterisation check. Begin by writing the rate with its unit and converting the requested interval to that same unit. The product of rate and time must be dimensionless and represents an expected count. For a Poisson count, state the interval and evaluate the probability for the requested integer.
For an exponential wait, define whether the variable measures time until the first event or residual time after an elapsed period. Connect the two views with the event that no arrival occurs during an interval. Practise this equivalence until switching between count and time language becomes automatic.
Build a superposition table for several independent sources: list each rate, sum them, and verify the combined process has the expected mean count over a test interval. Then reverse the operation through independent thinning. Multiply the original rate by each class probability and check that class rates add back to the original.
State why classification is independent; congestion-dependent or history-dependent routing does not inherit the simple thinning result. Use competing exponential clocks by summing their rates for the minimum time and dividing each rate by the total for the probability that its event occurs first. Check that those probabilities sum to one and that the faster clock is more likely to win. Keep process assumptions visible.
Stationary increments rule out a changing rate over time, independent increments rule out serial dependence between disjoint intervals, and the small-interval behaviour rules out substantial batch arrivals. Test plausibility with equal-window counts, the mean-to-variance comparison, lag correlation and a plot across time. Explain why aggregated call or flow arrivals may be closer to Poisson than bursty packet traffic.
On the handwritten page, place the count formula, exponential survival, superposition, thinning and competing-clock results around one unit check. In a final answer, identify the interval, expected count and mechanism that supports the model instead of presenting a probability without context.
Working through Poisson Processes and Exponential Times in EE8405? Sia is AskSia’s AI Electrical Engineering tutor — ask any EE8405 Poisson Processes and Exponential Times question and get a clear, step-by-step explanation grounded in how EE8405 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.
