Sorry, you need to enable JavaScript to visit this website.

Automatic Modulation Classification (AMC) has received a major attention last decades, as a required step between signal detection and demodulation. In the fully-blind scenario, this task turns out to be quite challenging, especially when the computational complexity and the robustness to uncertainty matter. AMC commonly relies on a preprocessor whose function is to estimate unknown parameters, filter the received signal and sample it in a suitable way. Any preprocessing error inherently leads to a performance loss.


The purpose of this note is to provide an effective means to compute the Cramér-Rao lower bound numerically for deterministic parameter estimation problems with the use of symbolic computation in MATLAB.


We consider the problem of estimating discrete self- exciting point process models from limited binary observations, where the history of the process serves as the covariate. We analyze the performance of two classes of estimators: l1-regularized maximum likelihood and greedy estimation for a discrete version of the Hawkes process and characterize the sampling tradeoffs required for stable recovery in the non-asymptotic regime. Our results extend those of compressed sensing for linear and generalized linear models with i.i.d.


This work is a part of our research on scalable and/or distributed fusion and sensor calibration. We address parameter estimation in multi-sensor state space models which underpins surveillance applications with sensor networks. The parameter likelihood of the problem involves centralised Bayesian filtering of multi-sensor data, which lacks scalability with the number of sensors and induces a large communication load. We propose separable likelihoods which approximate the centralised likelihood with single sensor filtering terms.


In \cite{Lomb}, Lomb developed a nonlinear regression approach to estimating the frequency of a noisy sinusoid when the measurement times were not equispaced, and a method for correcting the times so that the resulting regression sum of squares appeared very similar to the usual periodogram.