arXiv subjects

Peter Jeavons

Publications and source records attributed to Peter Jeavons.

5 recordsLinked to original sources

arXiv2018

Deep Learning Classification of 3.5 GHz Band Spectrograms with Applications to Spectrum Sensing

In the United States, the Federal Communications Commission has adopted rules permitting commercial wireless networks to share spectrum with federal incumbents in the 3.5~GHz Citizens Broadband Radio Service band. These rules require commercial systems to vacate the band when sensors detect radars operated by the U.S. military; a key example being the SPN-43 air traffic control radar. Such sensors require highly-accurate detection algorithms to meet their operating requirements. In this paper, using a library of over 14,000 3.5~GHz band spectrograms collected by a recent measurement campaign, we investigate the performance of thirteen methods for SPN-43 radar detection. Namely, we compare classical methods from signal detection theory and machine learning to several deep learning architectures. We demonstrate that machine learning algorithms appreciably outperform classical signal detection methods. Specifically, we find that a three-layer convolutional neural network offers a superior tradeoff between accuracy and computational complexity. Last, we apply this three-layer network to generate descriptive statistics for the full 3.5~GHz spectrogram library. Our findings highlight potential weaknesses of classical methods and strengths of modern machine learning algorithms for radar detection in the 3.5~GHz band.

eess.SP↗

arXiv2016

Feedback from Nature: Simple Randomised Distributed Algorithms for Maximal Independent Set Selection and Greedy Colouring

We propose distributed algorithms for two well-established problems that operate efficiently under extremely harsh conditions. Our algorithms achieve state-of-the-art performance in a simple and novel way. Our algorithm for maximal independent set selection operates on a network of identical anonymous processors. The processor at each node has no prior information about the network. At each time step, each node can only broadcast a single bit to all its neighbours, or remain silent. Each node can detect whether one or more neighbours have broadcast, but cannot tell how many of its neighbours have broadcast, or which ones. We build on recent work of Afek et al. which was inspired by studying the development of a network of cells in the fruit fly~\cite{Afek2011a}. However we incorporate for the first time another important feature of the biological system: varying the probability value used at each node based on local feedback from neighbouring nodes. Given any $n$-node network, our algorithm achieves the optimal expected time complexity of $O(\log n)$ rounds and the optimal expected message complexity of $O(1)$ single-bit messages broadcast by each node.We also show that the previous approach, without feedback, cannot achieve better than $Ω(\log^2 n)$ expected time complexity, whatever global scheme is used to choose the probabilities. Our algorithm for distributed greedy colouring works under similar harsh conditions: each identical node has no prior information about the network, can only broadcast a single message to all neighbours at each time step representing a desired colour, and can only detect whether at least one neighbour has broadcast each colour value. We show that our algorithm has an expected time complexity of $O(Δ+\log n)$, where $Δ$ is the maximum degree of the network, and expected message complexity of $O(1)$ messages broadcast by each node.

cs.DC↗

arXiv2014

Local Consistency and SAT-Solvers

Local consistency techniques such as k-consistency are a key component of specialised solvers for constraint satisfaction problems. In this paper we show that the power of using k-consistency techniques on a constraint satisfaction problem is precisely captured by using a particular inference rule, which we call negative-hyper-resolution, on the standard direct encoding of the problem into Boolean clauses. We also show that current clause-learning SAT-solvers will discover in expected polynomial time any inconsistency that can be deduced from a given set of clauses using negative-hyper-resolvents of a fixed size. We combine these two results to show that, without being explicitly designed to do so, current clause-learning SAT-solvers efficiently simulate k-consistency techniques, for all fixed values of k. We then give some experimental results to show that this feature allows clause-learning SAT-solvers to efficiently solve certain families of constraint problems which are challenging for conventional constraint-programming solvers.

cs.AI↗

arXiv2013

A Markov chain model of evolution in asexually reproducing populations: insight and analytical tractability in the evolutionary process

The evolutionary process has been modelled in many ways using both stochastic and deterministic models. We develop an algebraic model of evolution in a population of asexually reproducing organisms in which we represent a stochastic walk in phenotype space, constrained to the edges of an underlying graph representing the genotype, with a time-homogeneous Markov Chain. We show its equivalence to a more standard, explicit stochastic model and show the algebraic model's superiority in computational efficiency. Because of this increase in efficiency, we offer the ability to simulate the evolution of much larger populations in more realistic genotype spaces. Further, we show how the algebraic properties of the Markov Chain model can give insight into the evolutionary process and allow for analysis using familiar linear algebraic methods.

q-bio.PE↗

arXiv2012

Feedback from nature: an optimal distributed algorithm for maximal independent set selection

Maximal Independent Set selection is a fundamental problem in distributed computing. A novel probabilistic algorithm for this problem has recently been proposed by Afek et al, inspired by the study of the way that developing cells in the fly become specialised. The algorithm they propose is simple and robust, but not as efficient as previous approaches: the expected time complexity is O(log^2 n). Here we first show that the approach of Afek et al cannot achieve better efficiency than this across all networks, no matter how the probability values are chosen. However, we then propose a new algorithm that incorporates another important feature of the biological system: adapting the probabilities used at each node based on local feedback from neighbouring nodes. Our new algorithm retains all the advantages of simplicity and robustness, but also achieves the optimal efficiency of O(log n) expected time.

cs.DC↗