AUTOMATED ENERGY SAVING (AES) PARADIGM TO SUPPORT

PEDAGOGICAL ACTIVITIES OVER WIRELESS SENSOR NETWORKS

Abdul Razaque1 and Khaled Elleithy1

University of Bridgeport 126 Park Avenue,

Bridgeport, CT 06604 USA

{arazaque,elleithy}@bridgeport.edu

Abstract. Fast expansion in ambient intelligence (AmI) has attracted different walks of people. AmI systems

provide robust communication in open, dynamic and heterogeneous environments. This paper presents a AES

paradigm that introduces wireless sensor networks to control remote servers or other devices at remote place through

mobile phones. The main focus of paper is to consume minimum energy for obtaining the objectives. To realize the

paradigm, mathematical model is formulated. The proposed paradigm consists of automatic energy saving model

senses the environment to activate either the passive or active mode of sensor nodes for saving energy. Simulations

are conducted to validate the proposed paradigm; we use two types of simulations: Test bed simulation is done to

check practical validity of proposed approach and Ns2 simulation is performed to simulate the behavior of wireless

sensors network with supporting mathematical model. The prototype can further be implemented to handle several

objects simultaneously in university and other organizations.

Keywords: BT sensor node, AES mathematical model, saving energy, Ubiquitous communication.

1. Introduction

The Scientific and technological inventions have been observed for many decades that change the lives. The faster

growth in smallness of microprocessors has made significant progress for AmI [1]. AmI nourishes from many well

organized fields of computing and engineering.

Many objects are now embedded with computing power like home appliances and portable devices (e.g.,

microwave ovens, programmable washing machines, robotic hovering machines, mobile phones and PDAs). These

devices help and guide us to and from our homes (e.g., fuel consumption, GPS navigation and car suspension) [2].

AI involves compact power that is adapted to achieve specific tasks. This widespread accessibility of resources

builds the technological layer for understanding of AI [6], [8].

Information and communications technologies (ICT) have highly been accepted as part of introducing new cost-

effective solutions to decrease the cost of pedagogical activities and healthcare. For example, the Ubiquitous

intelligence health home that is equipped with AmI to support people in their homes. While this notion had some

problems to be fully understood in the past, but due to emerging technologies and incredible progress in low-power

electronics and sensor technologies supported with faster wireless network have facilitated the human life. Robust

heterogeneous wireless sensor network systems can be organized for controlling mobility of objects and logistics

[17].

These developments have led to the introduction of small-sized sensors that are capable for monitoring the

constraints of humans as well as living environment [3]. The objective of AmI with use of sensors is to provide

better living standard [19].

The deployment of mobile devices as sensor node provides more flexibility to interact with objects in any environment

[4]. These solutions not only improve the quality of education and health of people in their own homes but also provide

the fastest way of communication to interact with devices all over the world. The exploitation of mobile devices in

wireless sensor network provides more flexibility, intelligence and adaptivity to interact with devices dynamically in any

environment [11]. This makes it possible to deploy mobile phone not only as terminal but also as remote controller for

several devices. With deployment of mobile infrastructure, larger area can be covered as compare with installed

infrastructure using same number of sensors [10]. From other side, wireless networks face many challenging issues

including unreliable communication, consumption of energy, storage resources, inadequate computing and harsh

environments.

In this paper, we introduce a novel paradigm that involves mobiles and sensors. In particular, our goal is to introduce

faster and robust wireless sensor network to facilitate in controlling remotely available servers and different devices.

1. Related Work and Background Study

Controlling remotely available servers using mobile devices is one of highly challenging issues because of

scalability, interoperatibility, limited services and security of mobile devices. The access of one or multiple servers

from remote places causes of saving the resources and fostering communication. The mobile devices supported with

sensors make the task quicker and smarter but from other side, sensors consume more energy. No matured paradigm

has been found in literature that may indicate that devices are being controlled remotely using the sensors and

mobiles selecting robust efficient path.

The salient features of most related work are discussed in following studies. Until now, adhoc solution has been

deployed in home automation with support of current technologies to fulfill the requirements of users. Java based

home automation system based on internet has been introduced [5]. Authors manage some digital input and output

lines which are connected with appliances. Internet based wireless home automation system for multifunctional

devices have been introduced [15] to provide remote access. The authors in [5] used radio frequency link and simple

management protocol to handle these devices.

Home automation solution for indoor ambient intelligence (AmI) has been implemented in [12]. Authors have

used gateway and local control panels to maintain security system for house. Security involves with communication

and performs several entities same time. The IP based communication platform has been developed to interact with

security company and security staff.

Authors in [13] introduced building automation systems to control house hold appliances with support of typical

services and standard applications models. The paper also focused on BACnet, EIB/KNX and Lonworks as open

system in building automation domain. Three-level working model was implanted inside automation pyramid that

reduced convolution of individual level and kept levels transparent and lean.

2. Proposed Architecture

Our proposed architecture consists of two types of devices: mobile phone and specific type of BT node sensors. The

mobile phone is used to interact with different types of servers and other devices. Initially mobile device is deployed

to control remote servers but it can further be implemented to control several types of devices. The BT node

provides self-directed prototyping platform based on microcontroller and Blue-tooth radio. It has been introduced as

exhibition platform for research to deploy in distributed sensor networks, wireless communication and ad-hoc

networks. It is composed of microcontroller, separate radio and ATmega 128. The radio of BT node comprises of

two radios: first is low power chipcon CC1000 suited for ISM-band broadcast. It works same as Berkeley MICA2

mote does. This supports to BT node to create multi-hop networks. Second radio is Zeevo ZV4002 supported with

Bluetooth module.

We deploy BT node sensors in wireless sensor network to control remote servers and objects to make faster

communication consuming less energy using mobile phones. The BT node provides multiple interfaces to control

many devices at the same time but issue is consumption of lot of energy during the sensing time. We implement

paradigm to enforce BT node to work either in active or passive mode for saving the energy.

We use different protocols and standards in our architecture. Most of sensors do not communicate with Zigbee/

IEEE 802.15.4 standard [18]. Wi-Fi and Bluetooth are also not compatible because both utilize unlicensed 2.4 GHZ

ISM bandwidth. So they are using same bandwidth and in resulting causes of interference between them. In

addition, both are also transmitting data at binary phase shift keying (BPSK) and quadric phase shift keying (QPSK).

Our selection of BT node sensors provides the compatibility to Zigbee, Bluetooth and Wi-Fi. It supports different

types of applications and having multitasking support. BT node consists of several drivers that are designed with

fixed length of buffers. The buffers in BT node are adjusted during compile time to fulfill severe memory

requirements.

Available drivers include real time clock, UART, memory, I2C, AD converter and power modes. We have

discovered several theories that enabled to make mobile phone compatible with BT node sensors over wireless

sensor networks. We install Asus WL-500GP that maintains IEEE 802.11 b/g/n standards that is equipped with USB

port to interlink with sensors. We also use Zigbee USB adapter/ IEEE 802.15.4 to provide the communication

between sensors. Zigbee/ IEEE 802.15.4 also provide the capability to sensors to maintain multi-hop

communication.

We deploy highly featured hybrid network shown in figure 1. This multi-featured network supports to all BT

node sensors and different product of mobile families supported with Zigbee/IEEE 802.15.4 and IEEE 802.11 b/g/n

standards.

The beauty of sensor network is distribution into different regions. Each region has one boundary node that

coordinates with boundary node of other regions. Participating sensors go automatically into active and passive

modes for saving the energy.

This process is also done with mathematical model for consuming less energy because only one node is activated

during communication process. Boundary node is dynamically selected that plays a role as communicator with other

regions. Network supports to mobiles to carry the commands in faster way to control remotely available servers and

other devices.

Figure 1: Proposed architecture for controlling remote servers and other devices

3. Automatic energy saving (AES) mathematical model

The detection process for indoor and outdoor environment (IOE) consists of K number of sensors that collect

information pertaining to IOE and makes decision Di. In process, the value of Di is set 1 for deciding presence of

indoor environment (IE) and otherwise Di =0, it means outdoor environment (OE). The detection process is based

on maximized probability of detection (PD), with respect to constraints on probability of unknown environment

(UE), and by Neyman-Pearson Lemma [16].

Thus, the IOE environment is obtained. With help of blinking observations, Neyman-Pearson Lemma explains

optimal detector design of detector but we map this to map for IOE environment. In other words, the sensor detects

environment with realization of random variables K, UE and PD. Since probability of unknown environment (UE) is

random variable, which explains constraint of optimized problem in form of UE= α is no more applicable. One

BT

node

dsnT

rev3

BT node

dsnT

rev3

sensor

AMBIANT

INTELLIGENCE

SUPPORTED

MOBILE

AMBIANT

INTELLIGENCE

SUPPORTED

COMPUTER

OFFICE NETWORK

UNIVERSITY NETWORK

INDUSTRIAL NETWORK

INTELLIGENCE HOME

NETWORK

INTELLIGENCE

SECURITY NETWORK

MULTI-FUNCTIONAL INTELLIGENCE

AMBIANTCONTROLLER

UBIQUITOUS COMPUTING

UBIQUITOUS COMMUNICATION

INTELLIGENT INTERFACE

CONTEXT AWARNESS

COMPONANTS OF AMBIENT

INTELLIGENCE (AI)

MOBILE SUPPORTED CONTENT SERVER

Internet

important selection for statistical optimized quantity is the expected value of UE and PD. Hence, we maximize

expected value of probability to detect UE, with respect to constraints of expected value of probability [12].

Max E (PD) and E (UE) = α

&

: Hence,

are linked by relative operating characteristics (ROC) curve, we assume the functional form of

relationship is according to f( . Thus

(2)

The Problem of optimization of constraint can be defined as follows:

Max E(PD) and E (UE) = α

0 ≤ ≤ 1, k = { 0, 1, 2,…,} Here, is vector of realization of unknown environment

…]

In realistic environment, the UE cannot be infinite, since probability of getting many number of samples of

environment, P ( k=k) during finite T, reaches 0 as K → , for any available environment.

Lagrangian multiplier method can be used to resolve these issues as follows:

α

α

Distinguishing with respect to , and equating to 0, So we obtain:

This can be further expressed as:

α

K is sensor node of N that is case of finite capacity environment, then equation (3) represents N equations in ,

and combine with equation (4). Thus, we can resolve N +1 by using , , where k ={ 0, 1, 2, 3…, N}. We can

obtain the threshhold of UE using inverse of relationship of . Therefore, we can obtain unknown environment

but we have also to find indoor and outdoor environment. We are using following probabilities for indoor and

outdoor environment.

UEi = P (Di =0 | IOE outdoor), i = P (Di0 = 0| IOE outdoor), PDi = P (Di = 1| IOE indoor), i = P (Di0 =1| IOE

indoor). Assume that detection of environment is independent made by sensor, the UE probability of obtained

decision from ith sensor is given by following equation:

Here Pc1(Pc0) is probability of IOE. When sensor ith transmits bit for finding “0” and “1”. For IOE, Pc1 = Pc0 =

Pc; Therefore, (5) can be simplified as follows:

The probability of detection the IOE depends on receiving decision can be obtained from (6) and replacing UE with

PD. FOR IOE

This helps to determine the IOE, on basis of decision, BT node automatically works either active or passive mode.

This automatic process of detection causes the saving energy. We replace with IOE.

are showing the nature of environment, ; we substitute with Di. We

get:

If we obtain the value of Di, replace probability detection (PD) with its substitute values.

; Now rearrange and get Di.

If we obtain the value Di =1, it means passive mode is initiated and sensor h saves the energy. Passive mode is

supported by energy of sun. Di=0 gives the sign of active mode, and BT node consumes the energy in this mode,

which obtains through passive mode. If Di ≥ 1 || Di ≤ 0, shows that environment is unknown and sensor h does not

work and goes to sleep position in order to save the energy.

We can obtain energy preserving as follows:

Where EN(X) and EN(Y) denote total energy used by two different networks. E(i,j) indicates the energy used by

node i and j during transmission. h

Assume that aij is different from 0, only if node j receives E(i,j) when node i transmits. This gives appropriate

minimum level Emin i.e. if

We incur from the equation (13), it consume less energy during the process.BT sensor node follows the energy

saving integration method during the passive process. We here show numerical time integrators that causes of

preserving energy P(e), we begin by assuming an x-point quadrature formula with nodes Ni. The required weight of

ai is obtained through Lagrange basis polynomials in interruption that is shown as follows:

Let a1, a2, a3,…ax be different real numbers (usually 0 ≤ Ni ≤1) for which ai ≠ for all i. we use polynomial p(d0)

for satisfying the degree.

formula with nodes Ni and weights ai decreases integrator to specific collection of methods. We use polynomial

degree 2x – 1, thus Gauss points Ni that is equal to 0 and shifted with Legendre polynomial specific collection for

A(x). This treats arguments in A(x) and with different way that is considered as partitioned numerical

method. The solution of these methods depends on specific factorization of vector filed. Assume, If A(x) = A is

constant matrix, let (1, 1) be Hamiltonian system, thus it becomes energy saving integrator of [9]. This is proof that

sensor node also consumes minimum amount of energy and saves the energy in passive mood.

4. Simulation Setup and Result

Real wireless sensor environments use low power radios and are known due to high asymmetrical communication

range and stochastic link attributes. The simulation results could be differing extensively from realistic experimental

results, if network simulator makes only simple assumptions on wireless radio propagation [14]. Exact simulation to

the features of real wireless radios with diverse transmission powers is significant for assessing the realistic

performance of any new introduced model and algorithm in wireless sensor network. For our experimental

simulation setup, we use ns-2.35 RC7. Paradigm provides baseline for topology because this model provides the

capability to sensors to sense the environment. On basis of sensing process, sensors are triggered either remain in

active or passive mode.

The wireless sensor network is distributed into different regions to make the sensors more convenient to collect

information quicker. We have already discussed about boundary node that is playing role as anchor point (AP) or

head node.

We have set one boundary node dynamically in each region. Boundary node forwards the collected information of

its region to next region. In our case, it is not necessary that boundary node may always coordinate with only

boundary node of other region but it can forward the gathered information at 1-hop destination using proactive video

patching routing protocol (PVPRP) that combines features of DSDV and WRP routing protocols [7]. We have

simulated two types of different scenarios: First scenario is test bed that only controls single server from remote

access and consisting of two BT node sensors. In this scenario, one sensor is connected with mobile device and other

sensor is connected with server, we do not examine the performance of used mathematical model in first scenario.

The goal of first scenario is to show practical validity of our approach.

Later scenario is done with only ns2. In later scenario, we control multiple servers using multiple mobiles from

remote access using ns2. This is congested scenario and all four regions, serving their jobs but we have tested the

performance of WSN shown in figure 2.

Our objective in this scenario is to interact with remote servers and other objects. The scenario is particularly

designed to examine the behavior of WSN implementing mathematical model. This scenario is real test of WSN. We

have deployed 70 sensors within network area of 160m × 160m. Area is divided into 40m x 40m regions.

Sensors are randomly located within each region. The sinks are located at different positions but first sink is

available on (140, 60) because multiple mobile devices are using WSN and getting remote access to several servers

and devices.

Figure 2: NAM screenshot of proposed Scenario

The bandwidth of node is 50 Kbps and maximum power consumption for each sensor is set 160 mW, 12 mW and

0.5 mW for communication, sensing and idle modes respectively. In our case there is no idle mode. Sensors either

go to active or sleep mode. Each sensor is capable of broadcasting the data at 10 power intensity ranging from -20

dBm to 12 dBm. Total simulation time is 35 minutes and there is no pause time during the simulation but we set 30

seconds for initialization of phase at beginning of simulation. During this phase, only sensors onto sink remain

active and remaining sensors of all regions go into power saving mode automatically. The results shown in this

section are average of 10 simulation runs.

4.1. Efficiency of Network

With objective of validating this novel environment of WSN for handling servers and other several devices, we

conduct several tests from different perspective. Having proved the mathematical model, let us now evaluate the

coverage efficiency of network. We have already discussed about area of network that is 160m × 160m and divided

into 40m x 40m regions. We target coverage efficiency of network after deploying from 1 to 70 sensors Figures 3

and 4 show coverage efficiency of WSN during whole simulation time and increased number of sensors.

CO

VE

RA

GE

EFF

ICIE

NC

Y O

F N

ETW

OR

K

0

0

0.2

0.4

0.6

0.8

1

10 20 30 40 50 60 70

TRADITIONAL

WSN

AES WSN

NUMBER OF SENSORS

CO

VE

RA

GE

EF

FIC

IEN

CY

OF

NE

TW

OR

K

0

0

0.2

0.4

0.6

0.8

1

5 10 15 20 25 30 35

AES WSN

TIME (MINUTES)

TRADITIONAL

WSN

Figure 3: Coverage efficiency on basis of sensors Figure 4: Coverage efficiency on basis of amount of time

Our simulated network shows that our proposed WSN achieves almost 100% efficiency whereas general wireless

sensor network gets 69% efficiency using 70 sensors. We establish 15 sessions simultaneously in order to determine

the actual behavior of network in highly congested environment. If we have less number of sensors, it is hard to

establish many sessions at the same time. So, it is very interesting direction of this research that 70 sensors can

provide path-connectivity for 15 mobile devices to interact with remotely placed devices at same time. In addition,

one mobile can interact with multiple devices at same time. Question is why to deploy more sensors in that area?

And logically answer is availability of several servers and devices at the different places.

More sensors are required to find path and provide the connectivity for enough number of mobiles working

concurrently. Trend of network coverage during the time of 35 minutes of simulation is 99.2% in our proposed

WSN whereas performance of general network is affected. This is another weakness of general WSN. Our proposed

method produces stable efficiency during all the simulation.

It is proved that duration of simulation either increases or decreases; do not affect the efficiency in our case. The

minimum number of sensors required for covering area can be calculated as: Nmin(s) is minimum number of

sensors to cover whole area to maintain connectivity and coverage. ‘r’ is sensing range of sensor and it should be

assumed that sensing range is smaller than dimensions of monitoring area.

is maximum number of sensors.

‘R’ is distance of total network. We prove this with help of lemma4.

Lemma 4:

is upper bound on R and Nmin(s) is lower bound on Si, where Nmin(s) =

.

Proof: Let upper bound be linear on R with maximum number of sensors (total number of sensors) Nmax(s)

whereas lower bound on Si is invariant with Nmax(s). In addition, these bounds are not considered tight as long as

they do not consider transmission radius ‘Tr’ of sensors. However, we need better heuristic solution to follow these

bounds closely if irrespective of changes occur in the parameters of network. Hence, life time of network should

linearly be with Nmax(s), and Si to be constant with Nmax(s).

5. Conclusion and Future Works

Novel WSN has been introduced to control remotely available servers and several types of devices through mobiles.

This is purely unique type of WSN that deploys BT node sensors to control the devices. We have introduced AES

paradigm that senses the environment to activate either passive or active mode of BT node sensors for saving

energy.

The advantage of this model is to activate only one sensor in whole region. When sensor finishes the job, it

automatically goes to sleep mode. To validate the proposed paradigm in WSN, we implement proposed

mathematical model in ns2.35-RC7. On the basis of findings, we prove that our proposed research have saved

maximum amount of energy as compare with traditional method of saving energy. In addition, we have achieved

objectives to control the remotely available servers and devices by consuming minimum energy resources.

In future, we will implement this simulation based work into testbed to control several devices simultaneously. In

addition, we will implement the applications of this contribution in ambient intelligence to control house automation

devices, office devices and several static and moving devices using mobile.

References

[1] J. Chin, V. Callaghan, and G. Clarke. Soft-appliances: A vision for user created networked appliances in digital

homes. Journal on Ambient Intelligence and Smart Environments (JAISE), 1(1):69–75, 2009.

[2] Diane J. Cook, Juan C. Augusto, and Vikramaditya R. Jakkula. Ambient intelligence: applications in society

and opportunities for artificial intelligence. Pervasive and Mobile Computing, 2009.

[3] H. Nakashima, A. Aghajan, and J.C. Augusto, editors. Handbook on Ambient Intelligence and Smart

Environments. Springer Verlag, 2009.

[4] S. Reddy, Vids Samanta, Jeff Burke, Deborah Estrin, Mark Hansen, Mani B Srivastava, MobiSense - Mobile

Network Services for Coordinated Participatory Sensing, Proc. of the 9th International Symposium on

Autonomous Decentralized Systems (ISADS), 2009.

[5] A.R,Al-Ali and M.Al-Rousan,“Java-Based Home Automation System,”IEEE Trans .Consumer Electronics,

vol..50, no.2, 2004, pp. 498–504.

[6] Chiu, D. K. W., Choi, S. P. M., Wang, M., & Kafeza, E, “ Towards Ubiquitous Communication Support for

Distance Education with Alert Management”, Educational Technology & Society, 11 (2), 92-106, 2008.

[7] Kajita, S., & Mase, K. (2006). uClassroom: Expanding Awareness in Classroom to Ubiquitous Teaching and

Learning. Proceedings of the 4th IEEE International Workshop on Wireless, Mobile and Ubiquitous

Technology in Education (pp.161-163), Los Alamitos: IEEE Computer Society.

[8] Milrad, M., & Spikol, D. (2007). Anytime, Anywhere Learning Supported by Smart Phones: Experiences and

Results from the MUSIS Project. Educational Technology & Society, 10 (4), 62-70.

[9] Antonio F. Gómez-Skarmeta, “An Integral and Networked Home Automation Solution for Indoor Ambient

Intelligence” IEEE Trans. on Pervasive computing, 2010, pp. 1536-1268.

[10] Zhang,Y., Partridge,K., Reich,J.2007.Localizing tags using mobile infrastructure. In Proceedings of the 3rd

International Symposium on location and context awareness (LoCA). Lecture notes in computer Science (LNCS

4718). Springer, (Oberpfaffenhofen, Germany, September 20-21, 2007), 279-296.

[11] Wu, C., Zhang, Y., Sheng, W., Kanchi, S.2010.Rigidity guided localization for mobile robotic sensor

networks. International. Journal of Ad Hoc and Ubiquitous Computing. 6 (2), 114-128.

[12] Waltenegus Dargie, Christian Poellabauer, “Fundamentals of wireless Sensor Networks”, Wiley Series on

Wireless Communications and Mobile Computing, 2010.

[13] Wolfgang Kastner, Georg Neubschwandtner Stefan Soucek and H. Michael Newman, “Communication

Systems for Building Automation and Control”, Proceedings of the IEEE, VOL. 93, NO. 6, JUNE 2005,

pp.1178-1203.

[14] Jerry Zhao and Ramesh Govindan, “Understanding packet delivery performance in dense wireless sensor

networks,” in Sensys), Los Angeles, CA, November 2003.

[15] A.Z.Alkar and U.Buhur,“An Internet Based Wireless Home Automation System for Multifunctional

Devices”, ”IEEE Trans. Consumer electronics ,vol.51,no.4,2005, pp. 1169–1174.

[16] Q. Fang, J. Gao, L. J. Guibas, V. de Silva, L. Zhang, “GLIDER: gradient landmark-based distributed

routing for sensor networks”, In Proceedings of the 24th Annual Joint Conference of the IEEE Computer and

Communications Societies (INFOCOM), Vol. 1 (2005), pp. 339-350.

[17] Vermesan.O, “Wireless smart systems beyond RFID” European Conference & Exhibition on integration

issues of miniaturized systems MEMS, MOEMS, ICs and electronic components, 9-10 April 2008, Barcelona,

Spain.

[18] P. Barontib, P. Pillaia, V. W.C. Chooka, S. Chessab, A. Gottab, Y. F.Hua, Wireless Sensor Networks: A

Survey on The State of The Art and the 802.15.4 and ZigBee Standards, Computer Communications, Volume

30, Issue 7, May 2007, pp. 1655-1695.

[19] L. Benini, E. Farella, and C. Guiducci, “Wireless sensor networks: Enabling technology for ambient

intelligence,” Micro electron Journal, vol. 37, No. 12, pp. 1639–1649, 2006.