This paper presents a Hidden Markov Model (HMM)-based speech enhancement method, aiming at reducing non-stationary noise from speech signals. The system is based on the assumption that the speech and the noise are additive and uncorrelated. Cepstral features are used to extract statistical information from both the speech and the noise. A-priori statistical information is collected from long training sequences into ergodic hidden Markov models. Given the ergodic models for the speech and the noise, a compensated speech-noise model is created by means of parallel model combination, using a log-normal approximation. During the compensation, the mean of every mixture in the speech and noise model is stored. The stored means are then used in the enhancement process to create the most likely speech and noise power spectral distributions using the forward algorithm combined with mixture probability. The distributions are used to generate a Wiener filter for every observation. The paper includes a performance evaluation of the speech enhancer for stationary as well as non-stationary noise environment.
标签： Telecommunications Processing Signal for
上传时间： 2020-06-01
上传用户：shancjb
In this thesis several asp ects of space-time pro cessing and equalization for wire- less communications are treated. We discuss several di?erent metho ds of improv- ing estimates of space-time channels, such as temp oral parametrization, spatial parametrization, reduced rank channel estimation, b o otstrap channel estimation, and joint estimation of an FIR channel and an AR noise mo del. In wireless commu- nication the signal is often sub ject to intersymb ol interference as well as interfer- ence from other users.
标签： Communications Space-Time Processing Wireless for
上传时间： 2020-06-01
上传用户：shancjb
Part I provides a compact survey on classical stochastic geometry models. The basic models defined in this part will be used and extended throughout the whole monograph, and in particular to SINR based models. Note however that these classical stochastic models can be used in a variety of contexts which go far beyond the modeling of wireless networks. Chapter 1 reviews the definition and basic properties of Poisson point processes in Euclidean space. We review key operations on Poisson point processes (thinning, superposition, displacement) as well as key formulas like Campbell’s formula. Chapter 2 is focused on properties of the spatial shot-noise process: its continuity properties, its Laplace transform, its moments etc. Both additive and max shot-noise processes are studied. Chapter 3 bears on coverage processes, and in particular on the Boolean model. Its basic coverage characteristics are reviewed. We also give a brief account of its percolation properties. Chapter 4 studies random tessellations; the main focus is on Poisson–Voronoi tessellations and cells. We also discuss various random objects associated with bivariate point processes such as the set of points of the first point process that fall in a Voronoi cell w.r.t. the second point process.
标签： Stochastic Geometry Networks Wireless Volume and
上传时间： 2020-06-01
上传用户：shancjb
A wireless communication network can be viewed as a collection of nodes, located in some domain, which can in turn be transmitters or receivers (depending on the network considered, nodes may be mobile users, base stations in a cellular network, access points of a WiFi mesh etc.). At a given time, several nodes transmit simultaneously, each toward its own receiver. Each transmitter–receiver pair requires its own wireless link. The signal received from the link transmitter may be jammed by the signals received from the other transmitters. Even in the simplest model where the signal power radiated from a point decays in an isotropic way with Euclidean distance, the geometry of the locations of the nodes plays a key role since it determines the signal to interference and noise ratio (SINR) at each receiver and hence the possibility of establishing simultaneously this collection of links at a given bit rate. The interference seen by a receiver is the sum of the signal powers received from all transmitters, except its own transmitter.
标签： Stochastic Geometry Networks Wireless Volume and II
上传时间： 2020-06-01
上传用户：shancjb
Under the Energy Independence and Security Act of 2007 (EISA), the National Institute of Standards and Technology (NIST) was assigned “primary responsibility to coordinate development of a framework that includes protocols and model standards for information management to achieve interoperability of Smart Grid devices and systems…” [EISA Section 1305]. 35 This responsibility comes at a time when the electric power grid and electric power industry are undergoing the most dramatic transformation in many decades. Very significant investments are being made by industry and the federal government to modernize the power grid. To realize the full benefits of these investments—and the continued investments forecast for the coming decades—there is a continued need to establish effective smart grid 36 standards and protocols for interoperability.
标签： Framework Roadmap NIST and
上传时间： 2020-06-07
上传用户：shancjb
Have you ever looked at some gadget and wondered how it really worked? Maybe it was a remote control boat, the system that controls an elevator, a vending machine, or an electronic toy? Or have you wanted to create your own robot or electronic signals for a model railroad, or per- haps you’d like to capture and analyze weather data over time? Where and how do you start?
标签： Introduction Workshop Hands-On Arduino
上传时间： 2020-06-09
上传用户：shancjb
This introductory chapter is devoted to reviewing the fundamental ideas of control from a multivariable point of view. In some cases, the mathematics and operations on systems (modelling, pole placement, etc.), as previously treated in introductory courses and textbooks, convey to the readers an un- realistic image of systems engineering. The simplifying assumptions, simple examples and “perfect” model set-up usually used in these scenarios present the control problem as a pure mathematical problem, sometimes losing the physical meaning of the involved concepts and operations. We try to empha- sise the engineering implication of some of these concepts and, before entering into a detailed treatment of the different topics, a general qualitative overview is provided in this chapter.
标签： MultivariableControlSystems
上传时间： 2020-06-10
上传用户：shancjb
Although state of the art in many typical machine learning tasks, deep learning algorithmsareverycostly interms ofenergyconsumption,duetotheirlargeamount of required computations and huge model sizes. Because of this, deep learning applications on battery-constrained wearables have only been possible through wireless connections with a resourceful cloud. This setup has several drawbacks. First, there are privacy concerns. Cloud computing requires users to share their raw data—images, video, locations, speech—with a remote system. Most users are not willing to do this. Second, the cloud-setup requires users to be connected all the time, which is unfeasible given current cellular coverage. Furthermore, real-time applications require low latency connections, which cannot be guaranteed using the current communication infrastructure. Finally, wireless connections are very inefficient—requiringtoo much energyper transferredbit for real-time data transfer on energy-constrained platforms.
标签： Embedded_Deep_Learning Algorithms
上传时间： 2020-06-10
上传用户：shancjb
General paradigm in solving a computer vision problem is to represent a raw image using a more informative vector called feature vector and train a classifier on top of feature vectors collected from training set. From classification perspective, there are several off-the-shelf methods such as gradient boosting, random forest and support vector machines that are able to accurately model nonlinear decision boundaries. Hence, solving a computer vision problem mainly depends on the feature extraction algorithm
标签： Convolutional Networks Neural Guide to
上传时间： 2020-06-10
上传用户：shancjb
Machinelearninghasgreatpotentialforimprovingproducts,processesandresearch.Butcomputers usually do not explain their predictions which is a barrier to the adoption of machine learning. This book is about making machine learning models and their decisions interpretable. After exploring the concepts of interpretability, you will learn about simple, interpretable models such as decision trees, decision rules and linear regression. Later chapters focus on general model- agnosticmethodsforinterpretingblackboxmodelslikefeatureimportanceandaccumulatedlocal effects and explaining individual predictions with Shapley values and LIME.
标签： interpretable-machine-learning
上传时间： 2020-06-10
上传用户：shancjb
