An information-theoretic projection design framework is proposed, of interest for feature design and compressive measurements. Both Gaussian and Poisson measurement models are considered. The gradient of a proposed information-theoretic metric (ITM) is derived, and a gradient-descent algorithm is applied in design; connections are made to the information bottleneck. The fundamental solution structure of such design is revealed in the case of a Gaussian measurement model and arbitrary input statistics. This new theoretical result reveals how ITM parameter settings impact the number of needed projection measurements, with this verified experimentally. The ITM achieves promising results on real data, for both signal recovery and classification.

The adaptive beamforming algorithm-minimum variance distortionless response (MVDR) has been studied based on the electronically steerable parasitic array radiator (ESPAR) antenna. The ESPAR antenna uses a single radio-frequency (RF) front end, and its beamforming is achieved by adjusting reactance loads of parasitic elements coupled to the central active element. In the proposed beamforming method, the MVDR beamformer optimizes weights applied to outputs of beams. The optimization problem is formulated as a second-order-cone programming (SOCP) problem including a Euclidean distance metric to approximate the optimal equivalent weight vector to a feasible solution. Then the ESPAR beampattern design strategy iterates between the SOCP problem and a simple projection of reactance loads. The simulations show that the proposed MVDR beamforming method based on an ESPAR antenna gives a beam steering at the desired direction and placing nulls at the interfering directions, and it converges fast. However, when the desired source is close to the interferer, the output signal-to-interference-plus-noise ratio (SINR) degrades and where we use the interference-plus-noise sample covariance matrix to improve the beamforming performance.

We consider design of linear projection measurements for a vector Poisson signal model. The projections are performed on the vector Poisson rate, $X\in\mathbb{R}_+^n$, and the observed data are a vector of counts, $Y\in\mathbb{Z}_+^m$. The projection matrix is designed by maximizing mutual information between $Y$ and $X$, $I(Y;X)$. When there is a latent class label $C\in\{1,\dots,L\}$ associated with $X$, we consider the mutual information with respect to $Y$ and $C$, $I(Y;C)$. New analytic expressions for the gradient of $I(Y;X)$ and $I(Y;C)$ are presented, with gradient performed with respect to the measurement matrix. Connections are made to the more widely studied Gaussian measurement model. Example results are presented for compressive topic modeling of a document corpora (word counting), and hyperspectral compressive sensing for chemical classification (photon counting).

Spectrum sensing for cognitive radio is studied, using an electronically steerable parasitic antenna receptor (ESPAR), which relies on a single RF front end, and therefore meets the demanding low cost, power and size requirements of modern wireless terminals. We develop a strategy whereby the angular domain is divided into sectors, that are accessed via beamforming on a time division basis, to detect signals from primary users. We study the performance of detection metrics based on energy received per beam, and also eigenvalue-based detection statistics through considering the covariance matrix across the various directional beams. The ESPAR is able to achieve over 6dBi SNR improvement due to its beamforming capability. Further, once primary users' signals have been detected, directional transmit opportunities which do not interfere with active PUs become available, and we develop an adaptive beamforming algorithm to capitalise on these to efficiently utilise the spatial domain, which numerically optimises the beampattern and antenna efficiency using a convex formulation. The resulting beampatterns give between -20dB and -30dB nulls in the primary user direction.

We formally define multi-capacity bin packing, a generalization of conventional bin packing, and develop an algorithm called Reordering Grouping Genetic Algorithm (RGGA) to assign VMs to servers. We first test RGGA on conventional bin packing problems and show that it yields excellent results but much more efficiently. We then generate a multi-constraint test set, and demonstrate the effectiveness of RGGA in this context. Lastly, we show the applicability of RGGA in its desired context by using it to develop an assignment of real virtual machines to servers.