For any , implies . Hence , and taking expected values gives the Chernoff bound
At this is the trivial bound one; an infinite exponential moment merely gives a noninformative bound. For positive at which all required moment-generating functions are finite, define . Independence of the source random variables implies
Applying the Chernoff bound to gives
Consequently
Here and, for a nontrivial target, .
The effective bandwidth is the capacity charge of one source for this exponential-moment tail criterion. Independent sources contribute additively. It incorporates variability as well as mean demand: the Jensen inequality gives , and, under exponential integrability near zero, as . For bounded traffic its large- limit is the essential peak, as described by the mean and peak limits of effective bandwidth. Thus larger gives more weight to rare high demands. This is a one-window demand bound; a queue's overflow probability over many windows requires additional temporal modeling, not just this single-window calculation.
For the normal distribution , completing the square in its density gives the moment-generating function
Write and for the aggregate mean and variance. The exponential-moment capacity condition becomes
For , the last two terms are minimized at , where their sum is . Therefore Gaussian effective bandwidth yields the sufficient condition
Equivalently, optimizing the exponential bound directly when gives ; when the best nonnegative- bound is one. The square-root condition is sufficient rather than necessary because an exponential bound does not equal the exact normal tail.
For the exact condition, independence and multiplication of the normal moment-generating functions show that . For , let be the standard normal distribution function. The normal law is continuous, so
Since is continuous and strictly increasing from zero to one, the exact Gaussian overflow constraint is
The constant depends only on the desired tail exponent , not on the source counts. It is positive when , zero when , and negative when . For example, at the exact normal quantile is about , whereas the Chernoff bound uses ; the latter reserves more capacity.
Finally, the positive-variance hypothesis must not be hidden. If , then almost surely, and the zero-variance boundary of an overflow constraint gives . For , the exact condition is then . In particular does not meet the target even though substituting into the printed non-strict square-root inequalities would allow it. The displayed normal-quantile equivalence applies to nondegenerate Gaussian traffic; the deterministic case requires this strict boundary convention.
For any , on the event one has . The Markov inequality therefore gives , including the trivial value one at . Taking the infimum proves the Chernoff bound.
For with finite moment-generating functions, independence gives
Consequently
An effective bandwidth is an exponential-moment measure of demand at a chosen tail parameter : independence makes these quantities additive. The spare capacity pays for the desired exponential tail bound. For a cumulative-demand process over time , the corresponding bandwidth would be ; here the time horizon is one. It is generally larger than mean demand because it charges for fluctuations, and mathematical optimization over selects the useful tradeoff.
For independent normal distributions, put
The Gaussian effective bandwidth is . For and , the sufficient condition becomes . Its left side is minimized at , giving
This is a sufficient Chernoff safety margin, not the exact normal tail quantile.
Indeed , so, writing for the standard normal distribution function,
The exact Gaussian chance constraint is therefore
This is necessary and sufficient when , even when is negative. The Chernoff coefficient is more conservative.
There is an important boundary qualification. If , then deterministically, and for the exact requirement is , not . For example, , , satisfies the printed square-root condition but has . Thus both Gaussian non-strict displayed forms require positive total variance, which follows if at least one flow is present and its variance is positive. With no positive variance, the deterministic boundary in an upper-tail chance constraint must be treated separately. If , the target upper bound is at least one and imposes no restriction; the displayed square-root discussion naturally assumes .