To control the lifetime and optimize the energy consumption of wireless sensor networks (WSNs), while ensuring that application needs are met, it is essential to dynamically reconfigure the network by adjusting the parameters of its nodes. A great deal of research has been devoted to proposing methods for reconfiguring WSNs. However, there are no comprehensive studies analysing the different reconfiguration parameters, architectures or strategies. For this reason, this paper proposes the first systematic literature review on reconfiguration mechanisms for self-organised wireless sensor networks (SOWSNs) and those controlled by an external controller, following the principles of software-defined wireless sensor networks (SDWSNs), with a particular focus on energy optimisation. The main objective of this work is to explore reconfiguration strategies in depth from three different aspects: the parameters adjusted to optimise energy (What?), the architectures used for decision making (Who?), and the moments of reconfiguration implementation (When?). In addition, this study identifies unexploited areas in this field and suggests that hybrid and predictive approaches could be a promising way of overcoming these gaps.

Gaussian processes (GPs) are popular nonparametric statistical models for learning unknown functions and quantifying the spatiotemporal uncertainty in data. Recent works have extended GPs to model scalar and vector quantities distributed over non-Euclidean domains, including smooth manifolds, appearing in numerous fields such as computer vision, dynamical systems, and neuroscience. However, these approaches assume that the manifold underlying the data is known, limiting their practical utility. We introduce RVGP, a generalisation of GPs for learning vector signals over latent Riemannian manifolds. Our method uses positional encoding with eigenfunctions of the connection Laplacian, associated with the tangent bundle, readily derived from common graph-based approximation of data. We demonstrate that RVGP possesses global regularity over the manifold, which allows it to super-resolve and inpaint vector fields while preserving singularities. Furthermore, we use RVGP to reconstruct high-density neural dynamics derived from low-density EEG recordings in healthy individuals and Alzheimer's patients. We show that vector field singularities are important disease markers and that their reconstruction leads to a comparable classification accuracy of disease states to high-density recordings. Thus, our method overcomes a significant practical limitation in experimental and clinical applications.