A Comparison of Algorithms for Detecting Synchronous Digital Devices Using Their Unintended Electromagnetic Emissions

Electronically initiated explosives can have unintended electromagnetic emissions, which propagate through walls and unshielded containers. These emissions, if properly characterized, can be used to quickly detect explosive threats. In this paper, an analytic model is developed for the unintended emissions of clocked digital devices, such as microcontrollers, which can be used as initiators. It is demonstrated that these emissions are clock-dependent, periodic train of impulses. An autoregressive model of these clock emissions is developed, and the model is validated using measurements of an 8051 microcontroller. Existing algorithms, including pitch-estimation and the epoch-folding algorithm, are surveyed for detecting generic digital devices with unknown clock frequencies and emissions characteristics. A novel detection algorithm, which uses pitch estimation, is proposed. The model is used, in a simulated environment, to evaluate the noise performance of the proposed algorithms. Results indicate that the pitch-estimation techniques are robust against jitter and have a 4-dB sensitivity improvement over epoch-folding algorithms.