Stefan Bosse - TOOL CHAIN FOR HARVESTING, SIMULATION AND MANAGEMENT OF ENERGY IN SENSORIAL MATERIALS

Special Issue Article

Journal of Intelligent Material Systems

and Structures

24(18) 2245–2254

Ó The Author(s) 2013

Reprints and permissions:

sagepub.co.uk/journalsPermissions.nav

DOI: 10.1177/1045389X13488248

jim.sagepub.com

Tool chain for harvesting, simulation

and management of energy in

Sensorial Materials

Thomas Behrmann

1,2

, Christoph Budelmann

2,3

, Stefan Bosse

2,4

,Dirk

Lehmhus

and Marc C Lemmel

1,2

Abstract

The continuing decrease in size and energy demand of electronic sensor circuits allows endowing engineering structures

and, to an increasing degree, materials with integrated sensing and data processing capabilities. Materials that adhere to

this description are designated as Sensorial Materials. Their development is multidisciplinary and requires knowledge

beyond materials science in fields like sensor science, computer science, energy harvesting, microsystems technology,

low-power electronics, energy management, and communication. Development of such materials will benefit from sys-

tematic support for bridging research area boundaries. The present article introduces the backbone of an easy-to-use

toolbox for layout of the energy supply of smart sensor nodes within a sensorial material. The fundamental approach is

transferred from rapid control development, where a comparable MATLAB/Simulink tool chain is already in use. The

main goal is to manage limited power resources without unacceptably compromising functionality in a given application

scenario. The toolbox allows analysis of the modeled system in terms of energy and power and allows analyzing factors

such as energy harvesting, use of predictive power estimation, power saving (e.g. sleep modes), model-based cognitive

data reduction methods, and energy aware algorithm switching. It is linked to a simulation environment allowing analysis

of energy demand and production in a specific application scenario. Its initial version presented here supports single self-

powered sensor nodes. A broad set of application cases is used to develop scenario-dependent solutions with minimum

energy needs and thus demonstrate the use of the toolbox and the associated development process. The initial test case

is a large-scale sensor network with optical fiber–based data and energy transmission, for which optimization of energy

consumption is attempted. The toolbox can be used to improve the power-aware design of sensor nodes on digital hard-

ware level using advanced high-level synthesis approaches and provides input for sensor node and sensor network level.

Keywords

Self-powered sensor nodes, energy harvesting, energy simulation, toolbox, smart energy management, sensor network,

optical sensor network, Sensorial Materials

Introduction

In recent years, microsystems technology and micro-

electronics have continuously shrunk the size of sensor

network components. This development has led to con-

cepts like smart dust, a vision of dust particle–sized

autonomous sensor nodes or motes incorporating com-

munication devices and able to organize themselves in

sensor networks when distributed in the environment—

effectively, cubic millimeter size was reached in 2001

(Warneke et al., 2001). Our own vision is about similar

motes embedded in materials, thus endowing them with

the ability to ‘‘feel’’: We define a sensorial material

as the combination of host material and motes (Lang

et al., 2011a, 2011b). There are multiple application

scenarios for such materials, ranging from ambient

intelligence to structural monitoring, the latter having

been discussed, for example, by Renton (2001) in the

same year Warneke et al. (2001) published their work.

Bremen Institute for Metrology, Automation and Quality Science

(BIMAQ), University of Bremen, Bremen, Germany

ISIS Sensorial Materials Scientific Centre, University of Bremen, Bremen,

Germany

Working Group Cyber-Physical Systems, German Research Center for

Artificial Intelligence (DFKI), Bremen, Germany

Working Group Robotics, Department of Computer Science, University

of Bremen, Bremen, Germany

Corresponding author:

Thomas Behrmann, Bremen Institute for Metrology, Automation and

Quality Science (BIMAQ), University of Bremen, Linzer Str. 13, 28213

Bremen, Germany.

Email: RT@bimaq.de

One of the central issues that need to be solved for both

smart dust and Sensorial Materials is energy and power

supply. In a way, this challenge is more complex for the

latter, since for smart dust, each mote usually has to be

capable of coping with lack or excess of energy indivi-

dually. In contrast, for Sensorial Materials, the prob-

lem can be addressed on different levels, for example,

in terms of local energy management, on the one hand,

and network-wide energy management and distribu-

tion, on the other hand. The example already illustrates

the many similarities to sensor network energy supply

in general. However, there are also major discrepancies.

For example, material-embedded sensor networks

imply that there is little room to remedy a faulty design,

as the system will not be accessible for addition of, say,

the extra battery later found out to be necessary. Thus,

Sensorial Materials require comprehensive develop-

ment tools to analyze their energetic situation through-

out their life cycle and under all conceivable service

conditions. It is the main concern of this study to sketch

the development and evaluation of a dedicated toolbox,

which combines all aspects of energy supply as back-

bone of such a development and at the same time as an

optimization environment. This integrative approach

responds to a major need, as is underlined, for example,

by the IDTechEx report on energy harvesting markets

2011, which predicts a considerable market growth,

while at the same time stating that ‘‘there is exciting

enabling technology but many component suppliers

sell horizontally when users want solutions, not compo-

nents’’ (Das, 2011). Our approach is to adapt rapid

control prototyping methodologies established in mea-

surement system layout for this purpose.

Every designer of measureme nt systems dealing with

battery-powered or self-powered systems has to consider

the energy behavior. Most classic measuring devices were

designed to work on constant power supply. Future

methods for Sensorial Materials are varying from low-

power design, adapting power consumption, and energy

management to self-organization, self-localization, fault

tolerance, cognition, and grid intelligence. These options,

however, are very often discussed on an individual basis

(Mathuna et al., 2008), while the basis for a comprehen-

sive methodology including all of them and specifically

also their interdependencies in relation to a realistic

application scenario is completely missing.

Known approaches for system simulation require deep

knowledge of the component’s physics and use domain-

specific simulation tool s like SPICE (Simulation Program

with Integrated Circuit Emphasis) or finite element (FE)

methods (Elvin and Elvin, 2009). Due to the complexity,

these models are not capable of analyzing long-term beha-

vior, and the components must be defined on early stage.

Other multiphysics analysis techniques develop their calcu-

lation formulas from scratch and get accurate results, but

have to invest high efforts (Liang et al., 1997; Liao and

Sodano, 2009; Sirohi and Mahadik, 2011).

A typical approach in system development is to cal-

culate the system’s power behavior and to assure that

the supply will never drop below the needed power of

the loads. This is often done by evaluating worst case

scenarios, like a dark cloudy day for solar-powered

devices. Adding safety factors will ensure that there is

always a surplus of power, but at the cost of greatly

oversizing the system for the specific application. In

borderline cases, feasibility is often judged negatively

by such calculations. In these cases, adapting the mea-

surement task to available power using energy-based

scheduling, adjustment of sample rate, or more sophis-

ticated adaptive calculation algorithms for leveled data

processing is preferable. Simulations of energy flows

then have to show that the systems will not fail in a rea-

listic environment. These simulation results are often

used to adjust the layout parameters of the system.

Experienced system designers will claim that an opti-

mal solution depends on the special circumstances

of the individual measurement task. Thus, for self-

powered measurement nodes, the system design cannot

be transferred directly from one application to another.

Thus, the toolbox represents a major support to the

layout of Sensorial Materials. It will consist of a simu-

lation toolkit for energy flows and a tool for designing

modular sensor systems with an emphasis on self-

powered systems.

The aim is to support the layout of interconnected

and interacting energy supply, conversion, storage, and

consumer components (including data processing). To

this end, the toolbox contains generic component

blocks implemented in MATLAB/Simulink. They

should be easy to use for a measurement systems

designer to layout single sensor nodes as well as sensor

meshes and networks of autonomous sensor nodes.

At a later stage, advanced methods and technologies

will be added like adaptive data processing, energy

management, and generation of specific hardware

design, completing the features of the toolbox. The

basic algorithms are developed on state-of-the-art low-

power microprocessor architectures. The development

road map of the toolbox ranges from microcontrollers

to programmable logic (field-programmable gate array

(FPGA)) and later on to special custom chips (applica-

tion-specific integrated circuit (ASIC)) (see Figure 1).

There are many possible scenarios for self-powered

sensor applications (Bartholmai and Ko

ppe, 2010;

Budelmann and Krieg-Bru

ckner, 2011; Moser, 2009).

One potential scenario that could be analyzed and

parameterized by the use of the proposed toolbox is

discussed in the following sections.

Sensor nodes

In the context of this article, a sensor node could realize

one or more measurements and could be supplied by

2246 Journal of Intelligent Material Systems and Structures 24(18)

single or multiple sources. Sensorial abilities should be

integrated in a single sensor node.

Figure 2 shows an example of implementing sensor

nodes. The elements of the model of a sensor node can

be identified on one side as the parts for data processing

like acquisition, computation, and communication. On

the other side, there is energy distribution like harvest-

ing, storage, and consumer. Though the research activi-

ties also focus on the optimization of data processing

and communication with respect to energy consump-

tion, this report concentrates on modeling, simulation,

and analysis of the energy branch.

For most self-powered applications, the energy pro-

vided by the energy harvesting device has to be

converted. In Figure 3, a more detailed view of a power

system for a sensor node is displayed. Power flow from

the triggered harvester is converted to higher voltage

levels and normally charges a small capacitor. When a

certain voltage level is reached, the following circuit

activates the main functions of the sensor node. A vol-

tage regulator is used to provide constant conditions

for the parts of the node like measurement unit, proces-

sor/microcontroller and, if needed, a radio transmitter.

Toolbox

The structure of the toolbox is in accordance with the

node scheme depicted in Figure 2 and ordered from the

energy’s point of view (Figure 4). For each element,

there are generic blocks in the structure to cover the

main functionality. Special parts can be derived and

added to the library.

A mask in Simulink can be provided as a graphical

user interface (GUI) for setting parameter values

Figure 1. Development plan of the tool chain.

COTS: components of the shelf; OS: operating system; HDL: Hardware description language; IP: intellectual property; FPGA: field-programmable gate

array; ASIC: application-specific integrated circuit.

Energy supply

Energy

conversion

Signal

processing

Energy storage

Energy distribution

Sensor with

energy harvesting capability

Receiver +

Transmitter

Control

Communication

Energy

harvesting

material

Sensor

Sensorial material

Power / Energy flow Information

Figure 2. Elements of a sensor node as a part of a sensorial

material with energy distribution.

Energy

Harvesting

Material

Rectifier and

Storage Capacitor

Sensing and

Switching

Voltage

Regulator

Transmitter

Excitation

P-Sensor

Micro-

controller

AC/DC or

DC/DC converter

Power / Energy flow Information

Figure 3. Example for a radio-based sensor node with devices

for energy harvesting and sensing.

AC: alternating current; DC: direct current.

Behrmann et al. 2247

comfortably, which is then connected to constant vari-

ables inside the block model implemented in Simulink.

All blocks of this toolbox are equipped with an input

mask for convenience, for example, to parameterize the

block according to the data sheet (see Figure 6). In the

following sections, some exemplary implementations

are described to show the basic structure of the

toolbox.

Energy sources

The energy sources group covers the different possibili-

ties for providing energy to the sensor system. In conven-

tional design, there is a constant source always providing

enough power. This tool enables the user to tailor the

ratings for average and maximum use. For the possibil-

ity of environment-dependent energy sources, the blocks

have the ability to be controlled by environmental condi-

tions, for example, by solar radiation in the case of a

photovoltaic cell. The scenario in section ‘ ‘Tool chain

application scenario’’ uses a solar cell generator (see

Figure 5), which converts light pulses into electrical

power P

MPP

using the photoelectric effect.

MPP

is calculated as follows (see Figure 6)

MPP

= h  A  E = FF  I

 U

= I

 U

where h is the efficiency, E is the irradiance, A is the

area of the cell, FF is the fill factor, I

is the short-

circuit current, U

is the open-circuit voltage, I

is the

optimum operating current, and U

is the optimum

operating voltage.

Energy converters

Components for converting and regulating energy flows

are technically mandatory. Focusing on the energy, the

most important fact is that these devices have electric

losses and so the loss of power has to be calculated.

Most modern devices like direct current DC-to-DC

converters and so on use switched power operation and

smoothing capacitors. A generic approach for calculat-

ing losses without implementing the real operation is

sufficient at the level of system definition and could be

recalculated when the hardware layout is fixed. This

approach also speeds up the simulation for, for exam-

ple, life-time assessment (Figure 7).

Energy Sources

+ Batteries

+ Ideal Sources

+ Inductive/Magnetic

+ Mechanical

+ Piezo Electric

+ Photo Electric

+ Thermal Electric

+ Transformators

Energy Converters

+ AC/DC

+ DC/DC

+ Voltage Regulators

+PWM

+ Amplifier

Energy Storage

+ Accumulators

+ Capacitors

Energy Consumers

+ Actuators

+ Microcontrollers

+ Sensors

+ Radio Modules

+ Illuminants

Figure 4. Structure of toolbox.

AC: alternating current; DC: direct current; PWM: pulse width

modulation.

Optical Fibre

Energy harvesting

Pharvest

Figure 5. Photo generator block.

Figure 6. Block parameters solar cell block.

2248 Journal of Intelligent Material Systems and Structures 24(18)

Energy storage

The generic capacitor block contains a simple model of

a capacitor. The equivalent circuit is shown in Figure 8.

An initial voltage across the capacitor can be entered in

the mask.

The values R, L, C, and R

have to be entered into

the block’s input mask. Alternatively, the dissipation

factor tan d and the corresponding frequency can be

entered instead, which can be measured easily. The

series resistance R represents effects of dielectric losses

and conductor resistance. The capacitor inductance is

represented by the series inductance L. The parallel

resistance R

models leakage current flow and self-

discharge.

Energy consumers

The consumer block provides generic constant power

consumption. It is a good way to estimate the compo-

nent’s energy demand using datasheet’s values. It is

useful for modeling components, which have not been

modeled in detail yet. The cyclic wake-up consumer

block (application examples in Figure 9) models the

power consumption due to the electrical resistance of

the switch when in ‘‘on’’ position and switching losses.

The power consumed is calculated as

loss

= t

switch

 f

pwm

 U  I + I

 R

where t

switch

is the time needed to switch between on

and off, f

pwm

is the switching frequency, U is the voltage

of the power source, I is the current at maximum vol-

tage,

I is the average current, and R

is the ohmic resis-

tance of the switch when in ‘‘on’’ position (Graovac and

rschel, 2009). The output port provides information

on power consumption.

Tool chain application scenario

A first field test is aimed at optimizing the energy man-

agement of small sensor nodes within the SMaLL proj-

ect developed by the DFKI and other partners at the

University of Bremen (Budelmann and Krieg-Bru

ckner,

2011). The electronic sensor nodes are interconnected

only by an optical fiber, which is used for data exchange

and energy transfer. Local energy supply or storage is

not necessary, making the sensor nodes completely

maintenance free and ideal for the integration into

materials.

Figure 10 shows the first prototype of the SMaLL

sensor node, which has several sensor units onboard to

explore the range of applications. It already offers

reduced power consumption using low-power hardware

and advanced energy saving sleep modes.

Most of the modeled components are dominated by

a sleep schedule executing a cyclic wake-up reducing

energy demand by an estimated duty cycle of 1:100

(Figure 11).

When in operation, the sensor node’s power balance

is strictly negative. In order to drive the system directly,

a 17 times (see Figure 12 at graph minimum at t = 0.1

s) stronger light harvester would be needed to match

the short-term power demand.

When introducing an energy storage device, the

overall energy trend is positive, proving that the origi-

nal power supply is sufficient (Figure 13).

P_in P_out

DC-to-DC converter

Psu

Figure 7. DC/DC step-up converter.

DC: direct current.

Figure 8. Equivalent circuit of a generic capacitor.

P_consumed

Microcontroller

EFM32G210F128

P_consumed

Cyclic wakeup Sensor

Accelerometer

Figure 9. Microcontroller and cyclic wake-up sensor block as

examples for cyclic wake-up consumers.

Figure 10. Prototype of the SMaLL sensor node (Budelmann

and Krieg-Bru¨ckner, 2011).

Behrmann et al. 2249

The application of the analysis tool has supplied

valuable information about the layout of the sensor

node’s power system by showing the dynamic energy

flows of the different components. These results and

models can be incorporated into a subsequent hard-

ware synthesis step.

Energy management on algorithmic and

chip level

Typically, energy management is performed by a cen-

tral controller in which a program is implemented

(Lagorse et al., 2010), with limited fault tolerance and

the requirement of a well-known environment world

model for energy sources, sinks, and storage. Energy

management in a network can additionally involve the

transfer of energy between network nodes.

System-on-Chip (SoC) hardware design using

advanced high-level synthesis approaches on higher

algorithmic level can improve energy management and

power efficiency based on the results from toolbox

analysis. Models and model parameters provided by

the toolbox can be used for algorithm design and hard-

ware synthesis.

P+

P- 1

P- 2

P- 3

P- 4

P- 5

P- 6

Pbalance

Power

Balance

Power

Balance

P_consumed

Optical receiver

Module

P_consumed

Optical Transmitter

Module

Optical Fiber

Energy harvesting

P_consumed

Microcontroller

EFM32G210F128

Integrator

Energy

Balance

0.001395

Display

P_consumed

Cyclic wake-up Sensor

Temperature

P_consumed

Cyclic wake-up Sensor

DMS

P_consumed

Cyclic wake-up Sensor

Accelerometer

P_in P_out

DC/DC

Step-up converter

Psupply

Pbalance

Pharvest

Figure 11. Power model of a multisensor node.

DC: direct current.

Figure 12. Power balance. Figure 13. Energy trajectory.

2250 Journal of Intelligent Material Systems and Structures 24(18)

Energy management on node and microchip level

Energy management can be performed first at runtime

and second at design time using application-specific

SoC design methodologies, contributing to low-power

systems on both algorithmic and technology level. The

proposed tool chain offers advanced capabilities for

energy management on algorithmic level by analyzing

algorithms regarding their impact on power consump-

tion. Smart energy management can be performed spa-

tially at runtime by a behavior-based or state-action–

driven selection from a set of different (implemented)

algorithms classified by their demand of computation

power and different qualities-of-service and temporally

by varying data processing rates.

Signal and control processing is modeled on an

abstract algorithmic level using signal flow diagrams, as

shown in Figure 14. These signal flow graphs derived

from the toolbox can be mapped to Petri Nets to enable

direct high-level synthesis of digital SoC circuits using a

multiprocessing architecture with the communicating

sequential process model on execution level and the

high-level synthesis framework ConPro (Bosse, 2011).

Power analysis using simulation techniques on gate

level provides input for the algorithmic selection during

runtime and improvement of energy management of

the system at design time leading to a closed-loop

design flow. Additionally, the signal flow approach

enables power management by varying the signal flow

rate parameters.

The signal flow diagram is first transformed into an

S/T Petri Net representation. Functional blocks are

mapped to transitions, and states represent data, which

is exchanged between those functional blocks. The par-

titioning of functional blocks to transitions of the net

can be performed at different compositions and com-

plexity levels. The signal flow diagram is partitioned

using complex blocks (merging low-level blocks like

multipliers and adders) to reduce communication com-

plexity (and data processing latency).

Sensor data (I) is acquired periodically and passed to

the data processing system. A token of the Petri Net is

equal to a data set of one computation processed by the

functional blocks in the signal flow. Functional blocks

can be placed in concurrent paths of the net.

The Petri Net is then used first to derive the commu-

nication architecture and second to determine an initial

configuration for the communication network.

Functional blocks with a feedback path require the

injection of initial tokens in the appropriate states (not

required in the example).

States of the net are mapped to buffered communi-

cation channels, and transitions are mapped to concur-

rently executing processes—each with sequential

instruction processing—using the ConPro program-

ming language (Bosse, 2011).

Forked states indicate concurrency in the Petri Net

flow. Exploring concurrency in signal flow diagrams

using Petri Nets reduces latency for the computation of

one data set. Also, pipelining can decrease latency of a

data set stream significantly, derived again from the

Petri Net representation.

The derived multiprocess programming model is

finally synthesized to a digital logic SoC using the high-

level synthesis, as shown in Figure 14. For simulation,

gate-level synthesis is performed with a standard logic

cell technology library. The resulting net-list is analyzed

with an event-driven simulator, calculating the overall

cell activity for each time unit, defined by terms of cell

output changes, enabling power optimization on algo-

rithmic level.

Methods from artificial intelligence (AI) can be used

to manage energy at runtime with dynamic parameter

adaptation and algorithmic selection based on the

results from previous algorithm analysis. AI methods

differ in complexity; thus, only few are suitable to be

embedded in microchips, like decision trees (Bosse and

Kirchner, 2012).

Management on network level

In a sensor network, energy management can take place

both locally on each sensor node and globally covering

the whole sensor network. Locally energy consumption

is minimized, but globally energy can be transferred

between nodes to increase system stability. Again, AI

methods can be used to manage energy globally in the

sensor network. Having the technical abilities to trans-

fer energy between nodes using communication chan-

nels (e.g. optically, like in the SMaLL sensor node), it is

possible to use active messaging to transfer energy from

good nodes having enough energy to bad nodes, requir-

ing energy.

Initially, the sensor network is a collection of inde-

pendent computing nodes. Interaction between nodes

can be implemented using agents to manage and distri-

bute information and energy. Agents are suitable for

implementation on microchip level (Bosse and Pantke,

2013), as shown in Figure 15.

An agent can be sent by a bad node to explore and

exploit the near neighborhood. The agent examines

sensor nodes during path travel or passing a region of

interest (perception) and decides to send agents holding

additional energy back to the original requesting node

(action). Additionally, a sensor node is also represented

by a node agent. The node and the energy management

agents must negotiate the energy request. The toolbox

analysis can provide required input for the global

energy management strategy performed by multiagent

systems, for example, retrieved by machine learning

methods based on analysis and experimental results.

Behrmann et al. 2251

Conclusion

The described tool chain is now capable of simulating

and analyzing simple scenarios for the parameteriza-

tion of self-powered sensor nodes. As it is meant as a

tool for a measurement systems designer outlining new

scenarios for measurement applications, its usage is

deliberately kept simple and generic. For a full simula-

tion of, for example, the electric circuit (SPICE) or

communication issues over wide-spread wireless sensor

networks, other tools are more powerful and accurate.

Their drawback is that they are limited in scope,

requiring, for example, an already defined hardware

or an exact specification of it. The tool chain can con-

tribute to the process of defining and designing the

right hardware configuration. The specific focus on

energy harvesting principles facilitates the process of

exploring the area of self-powered sensors and

Sensorial Materials. The tool chain can improve and

support the hardware design of sensor nodes and the

design of energy management algorithms including

smart energy management strategies on sensor node

and network level.

Figure 14. Power-efficient system-on-chip hardware design using high-level synthesis and smart energy management on node level

using the tool chain in a feedback loop.

2252 Journal of Intelligent Material Systems and Structures 24(18)

Outlook

As a next step, the fundamentals of the realization of

the tool chain will be published together with tutorials

and comprehensive sample scenarios. In parallel, a

platform for scientific discussion on performance and

implementation of the modules will be set up. This will

provide the basis for future versions incorporating con-

tributions from the scientific community. The final aim

is a powerful, yet living tool for the measurement sys-

tems and Sensorial Materials designer alike, who may

be experienced in materials science, power systems,

energy management, computer science, and communi-

cations, but usually not an expert in all of them and

will thus profit from contributions of experts in com-

plementary fields. Additionally, the tool chain will

cover data processing, energy management, and low-

level communication on microchip level. From this

basis, automated generation of code for microcontrol-

lers and configurable hardware will enable rapid sensor

system prototyping following a development process

similar to rapid control prototyping. The tool chain

will then span the full range of energy-related system

Figure 15. Design of smart energy management algorithms with multiagent systems on network level using the tool chain in a

feedback loop.

Behrmann et al. 2253

components and functions from sensor signal capture

and conditioning via data evaluation and energy harvest-

ing to communication, and it will allow testing of various

combinations under realistic conditions as well as finding

optimum solutions for given boundary conditions.

Declaration of conflicting interests

The authors declare that there is no conflict of interest.

Funding

This study was supported by the State and University of

Bremen in the framework of the Integrated Solutions in

Sensorial Structure Engineering (ISIS) Sensorial Materials

Scientific Centre.

References

Bartholmai M and Ko

ppe E (2010) Funksensornetzwerk zur

Strukturu

berwachung und Schadensfru

herkennung an

Bauwerken. In: Puente Leon F, Sommer K-D and Heiz-

mann M (eds) Verteilte Messsysteme. Karlsruhe: Universi-

ts-Verlag Karlsruhe, pp. 49–56.

Bosse S (2011) Hardware-software-co-design of parallel and

distributed systems using a unique behavioural program-

ming and multi-process model with high-level synthesis.

In: Proceedings of the SPIE microtechnologies 2011 confer-

ence, Prague, Czech Republic, 18–20 April 2011, Session

EMT 102 VLSI Circuits and Systems.

Bosse S and Kirchner F (2012) Smart energy management

and energy distribution in decentralized self-powered sen-

sor networks using artificial intelligence concepts. In: Pro-

ceedings of the smart systems integration conference 2012,

session 4,Zu

rich, Switzerland, 21–22 March 2012.

Bosse S and Pantke F (2013) Distributed computing and reli-

able communication in sensor networks using multi-agent

systems. Prod Eng Res Dev 7: 43–51.

Budelmann C and Krieg-Bru

ckner B (2011) Sensor network

based on fibre optics for intelligent Sensorial Materials. In:

European conference and exhibition on advanced materials

and processes (EUROMAT 2011), Montpellier, France,

12–15 September 2011.

Das R (2011) Energy Harvesting Markets Analysed: Creating

a $4.4 Billion Market in 2021: Executive Summary & Con-

clusions. Cambridge, MA: IDTechEx, Inc.

Elvin NG and Elvin AA (2009) A coupled finite element-

circuit simulation model for analyzing piezoelectric energy

generators. J Intel Mat Syst Str 20(5): 587–595.

Graovac D and Pu

rschel M (2009) IGBT Power Losses Calcu-

lation Using the Data-Sheet Parameters. Infineon Applica-

tion Note V1.1. Neubiberg: Infineon Technologies AG.

Lagorse J, Paire D and Miraoui A (2010) A multi-agent sys-

tem for energy management of distributed power sources.

J Renew Energ 35(1): 174–182.

Lang W, Jakobs F, Tolstosheeva E, et al. (2011a) From

embedded sensors to Sensorial Materials—the road to

function scale integration. Sensor Actuat A: Phys 171(1):

3–11.

Lang W, Lehmhus D, van der Zwaag S, et al. (2011b) Sensorial

Materials—a vision about where progress in sensor integra-

tion may lead to. Sensor Actuat A: Phys 171(1): 1–2.

Liang C, Sun FP and Rogers CA (1997) Coupled electro-

mechanical analysis of adaptive material systems—

determination of the actuator power consumption and sys-

tem energy transfer. J Intel Mat Syst Str 8: 335–343.

Liao Y and Sodano HA (2009) Structural effects and energy

conversion efficiency of power harvesting. J Intel Mat Syst

Str 20: 505–514.

Mathuna CO, O’Donnell T, Martinez-Cartala RV, et al.

(2008) Energy scavenging for long-term deployable wire-

less sensor networks. Talanta 75(3): 613–623.

Moser C (2009) Power Management in Energy Harvesting

Embedded Systems. Aachen: Shaker.

Renton WJ (2001) Aerospace and structures: where are we

headed? Int J Solids Struct 38(19): 3309–3319.

Sirohi J and Mahadik R (2011) Piezoelectric wind energy har-

vester for low-power sensors. J Intel Mat Syst Str 22:

2215–2228.

Warneke B, Last M, Liebowitz B, et al. (2001) Smart dust:

communicating with a cubic-millimeter computer. IEEE

Comput 34(1): 44–51.

2254 Journal of Intelligent Material Systems and Structures 24(18)