Interference issues in modern communications systems.

摘要

A critical component of a communication system's design is the analysis of potential electromagnetic interference from within the system and from outside sources. Through physical and mathematical modeling, we can quantify the impact of interference on key performance metrics of the system and pursue an optimal design based on certain interference constraints. In this dissertation, we investigate several interference related issues in three emerging technologies: Broadband over Power Line (BPL), Dynamic Spectrum Access (DSA), and Mobile Ad Hoc Networks (MANET).;In the BPL study, we analyze the radio interference from a BPL system operating between 2 MHz and several tens of MHz. An overhead medium-voltage power line is modeled as a 3-phase set of parallel wires above a lossy earth. Both a near-exact solution and a closed-form far-field approximation are presented. The maximum allowable excitation voltage vs. frequency is computed by assuming compliance with FCC field strength limits. These calibration results are used to study the interference to both terrestrial and airborne services, using noise floor increase as a metric of interference severity. We also quantify the relationship between BPL capacity and BPL interference.;In the DSA study, we propose a solution to spectrum policy enforcement in DSA networks involving the detection of unauthorized spectrum usage. We formulate the anomalous usage detection problem using statistical significance testing. The detection problem is investigated considering two cases, characterized by whether the authorized (primary) transmitter is mobile or fixed. We propose a detection scheme for each case, respectively, by exploiting the spatial pattern of received signal energy across a network of sensors. Analytical models are formulated when the distribution of the energy measurements is given and we present an algorithm using machine learning techniques to solve the general case when the statistics of the energy measurements are unknown.;In theMANET study, we propose a two-phase interference classification framework in a CSMA/CA-based MANET. It classifies different jamming attacks in a 3-D metric space and distinguishes unintentional interference and interference-free conditions based on the consistency of ACK errors and received signal strength.