Skip to main content
Disciplines
Automotive Business IT + Informatics Construction + Real Estate Electrical Engineering + Electronics Energy + Sustainability Insurance + Risk Finance + Banking Management + Leadership Marketing + Sales Mechanical Engineering + Materials
Events
DE
EN
Books
Journals
Topic Page
Marketing
Start single access now
Access for companies
EXTENDED SEARCH
Log in
Top

2014 | Book

Introduction Read chapter Read first chapter

Intelligence for Embedded Systems

A Methodological Approach

Author: Cesare Alippi

Publisher: Springer International Publishing

Part of: Springer Professional "Wirtschaft+Technik" , Springer Professional "Technik" , Springer Professional "Wirtschaft"

Login to get access
insite
SEARCH

About this book

Addressing current issues of which any engineer or computer scientist should be aware, this monograph is a response to the need to adopt a new computational paradigm as the methodological basis for designing pervasive embedded systems with sensor capabilities. The requirements of this paradigm are to control complexity, to limit cost and energy consumption and to provide adaptation and cognition abilities allowing the embedded system to interact proactively with the real world. The quest for such intelligence requires the formalization of a new generation of intelligent systems able to exploit advances in digital architectures and in sensing technologies. The book sheds light on the theory behind intelligence for embedded systems with specific focus on:

· robustness (the robustness of a computational flow and its evaluation);

· intelligence (how to mimic the adaptation and cognition abilities of the human brain),

· the capacity to learn in non-stationary and evolving environments by detecting changes and reacting accordingly; and

· a new paradigm that, by accepting results that are correct in probability, allows the complexity of the embedded application the be kept under control.

Theories, concepts and methods are provided to motivate researchers in this exciting and timely interdisciplinary area. Applications such as porting a neural network from a high-precision platform to a digital embedded system and evaluatin

g its robustness level are described. Examples show how the methodology introduced can be adopted in the case of cyber-physical systems to manage the interaction between embedded devices and physical world.

Researchers and graduate students in computer science and various engineering-related disciplines will find the methods and approaches propounded in Intelligence for Embedded Systems of great interest. The book will also be an important resource for practitioners working on embedded systems and applications.

Table of Contents

Frontmatter
Chapter 1. Introduction
Abstract
The emergence of nontrivial embedded sensor units (e.g., those embedded in smartphones and other everyday appliances), networked embedded systems, and sensor/actuator networks (e.g., those associated with services running on mobiles and wireless sensor networks) has made possible the design and implementation of several sophisticated applications where large amounts of real-time data are collected to constitute a big data picture as time passes. Acquired data are then processed at local, cluster-of-units or server level to take the appropriate actions or make the most appropriate decision.
Cesare Alippi
Chapter 2. From Metrology to Digital Data
Abstract
Most embedded systems, and surely their aggregation in sensor networks, are characterized by applications that take advantage of the available rich sensor platform. However, we should not forget that uncertainty affects the information content carried by sensor data when designing the application. The chapter introduces the basic concepts behind measure and measurements, e.g., accuracy, resolution, and precision and sheds light on the elements composing the measurement chain (transducer, conditioning stage, analog to digital converter, estimation module). Since measurements are affected by uncertainty, we need to investigate how uncertainty corrupts the final acquired data. This analysis sets the basis for the subsequent propagation of uncertainty within any computational chain as well as introduces constraints on accuracy for the final embedded application.
Cesare Alippi
Chapter 3. Uncertainty, Information, and Learning Mechanisms
Abstract
This chapter introduces and formalizes the most important forms of uncertainty that digital embedded systems have to deal with and how it propagates within a computational flow. We found uncertainty associated with data representation of input variables (e.g., introduced by the truncation and rounding operators) and that propagated along the computational flow. Moreover, we have uncertainty associated with models learned from data and, finally, that introduced at the application design phase. All these sources of uncertainty combine in a nonlinear way and influence the result of the computation in execution on the embedded system, which becomes approximated. The main elements of the statistical theory of learning are then presented. It is shown how information can be extracted from noisy data and how the limited number of available training data, the effectiveness of the learning mechanism, and the uncertainty intrinsic with the problem affect the performance of the learned model.
Cesare Alippi
Chapter 4. Randomized Algorithms
Abstract
Every time we cannot solve a complex problem, either because it is too complex or computationally hard, we try to explore it on a number of instances and expect that, asymptotically with the number of samples, something can be said. This chapter introduces a key mechanism every engineer/computer scientist should be aware of, that of randomization. After having introduced the main results coming from the theory of randomization and those granting convergence of sampled quantities to the exact ones, we will present some general methods based on randomized algorithms for solving a large class of performance assessment problems. Results, independent of the dimension of the sampling space and the particular probability density function associated with sample drawing, hold in probability at arbitrary accuracy and confidence levels, function of the envisaged number of samples.
Cesare Alippi
Chapter 5. Robustness Analysis
Abstract
Robustness analysis deals with the problem of investigating whether a given solution is able to tolerate the presence of uncertainty/perturbations affecting the structural parameters or not. Current robustness analyses mostly assume the small perturbation hypothesis to make the mathematics amenable and derive the relationship between perturbations affecting the computational flow/variables and the induced loss in performance in a closed form. However, such results prove to be of limited use when we wish to port an algorithm to an embedded system, since the small perturbation hypothesis is mostly violated. This chapter solves this problem by introducing the perturbation in the large analysis: based on randomized algorithms we do not require any more the perturbation to be small in magnitude, hence, yielding effective estimates for the robustness index possessed by the application.
Cesare Alippi
Chapter 6. Emotional Cognitive Mechanisms for Embedded Systems
Abstract
The chapter provides an engineering-oriented perspective of the emotional processing exposed by the human brain. In fact, it is strongly believed that the next generation of embedded systems will natively integrate similar forms of intelligence in the device/application to provide sophisticated behaviors. Automatic and controlled conscious mechanisms are presented as well as the principal components taking part to the management of emotions. At the lower level of emotion processing we have fast automatic mechanisms, which show to be very effective but introduce possible erroneous evaluations. The conscious upper level, once activated, might overwrite decisions taken at the lower level by executing more complex functions that, following a false positive might even request reconfiguration of the functional elements composing the automatic level.
Cesare Alippi
Chapter 7. Performance Estimation and Probably Approximately Correct Computation
Abstract
The analysis phase of a problem aims at evaluating, given a computation, its performance. Performance can be intended in several ways depending on the specific target problem as well as the abstraction level where it is carried out. For instance, at the device level we have cost, latency, throughput, power, energy, and complexity to name some major performance design indexes. At the algorithm level we have accuracy, confidence, energy, and complexity. Not rarely we constrain such indexes and we saw in Chap. 4 how it is possible to evaluate their satisfaction level.
Cesare Alippi
Chapter 8. Intelligent Mechanisms in Embedded Systems
Abstract
The chapter investigates how intelligence plays a key role in managing basic mechanisms involving embedded systems and applications. At the lowest abstraction level we encounter those forms of intelligence that, through adaptation, affect the voltage/clock frequency of the system to reduce the power consumption as well as those aiming to maximize the harvested energy in photovoltaic cells. Adaptive sensing, clock synchronization, and intelligent localization of distributed embedded units are some basic problems that can be suitably addressed by considering automatic and conscious intelligent mechanisms. Functional reprogrammability both at the hardware and software level is another form of intelligence that allows the embedded application to undergo changes whenever needed also in remotely deployed units.
Cesare Alippi
Chapter 9. Learning in Nonstationary and Evolving Environments
Abstract
Previous chapters have developed methods and methodologies for solving specific aspects involving intelligent processing on embedded systems and presented techniques for their performance assessment. However, if we look carefully at those methods, we can observe that we have commonly assumed that the process generating the data acquired by our sensors was not changing with time (stationarity or time invariance assumption).
Cesare Alippi
Chapter 10. Fault Diagnosis Systems
Abstract
Sensors and real apparatus are prone to faults that, in turn, affect the quality of retrieved data. Detection of faults or erroneous behaviors in sensor data stream must be anticipated to prevent drastic side effects (recall we make decisions out of incoming data). Cognitive fault diagnosis systems aim at detecting, identifying, and isolating the occurrence of faults without assuming that the process generating the data is known. It is shown that little can be done at the single sensor level unless strong hypotheses are made. However, the situation is different if the embedded system mounts a rich sensor platform or is inserted in a sensor network. In such a case, redundancy in the information content and functional dependencies among sensors can be exploited to classify a change as fault, a change in the environment or an inefficiency of the change detection method (model bias).
Cesare Alippi
Backmatter
Metadata
Title
Intelligence for Embedded Systems
Author
Cesare Alippi
Copyright Year
2014
Electronic ISBN
978-3-319-05278-6
Print ISBN
978-3-319-05277-9
DOI
https://doi.org/10.1007/978-3-319-05278-6