Abstract: A method and apparatus for detecting transient episodes includes parallel connected transient episode detectors (peripheral devices) to measure short-lived disturbances concurrently, each uniquely associated in series with one of a multiplicity of analog sensors. A shared memory connects an object sensor with its uniquely associated transient episode detector. The transient episode detector outputs signals through a second shared memory to a host disturbance processor. The transient episode detector adaptively tracks and masks background noise through use of a noise statistics stack. The stack is operated in two different modes: initialization and update. In the initialization mode, the stack operates in a recursive sort from the bottom up to replace initial dummy data with ordered noise level peak value measurements. In the update mode, only new significant non-redundant noise level peak values which fall outside of a defined central normal range of the stack are used to update the stack.