Wireless Network Deployment as Low Cost Building Management

System Solution

Liam Moore

, Mike Hayes

, Brendan O. Flynn

, Cian O. Mathuna

, Emmanuel Frecon

Joakim Ericksson

, Peeter Kool

, Peter Rosengren

, Alberto Fernandez

, Jacek Rosik

and Donagh MacSuibhne

The Tyndall National Institute, Cork City, Ireland

The Swedish Institute of Computer Science, Stockholm, Sweden

CNET, Stockholm, Sweden

Sensing and Control, Barcelona, Spain

ResourceKraft., Limerick, Ireland

ARUP, Cork, Ireland

Keywords: Wireless Sensor Network, 6LoWPAN, BMS, Retrofit, Middleware.

Abstract: This paper presents the design and implementation of a wireless monitoring and actuation network for

residential and commercial buildings that was carried out as part of the ARTEMIS funded project ME3gas.

The aim of this deployment is to demonstrate that low cost wireless sensor networks can be used in

situations where a full building management system may not be suitable technically or commercially either

in residential home applications or commercial enterprises. This work focuses not just on electricity

consumption but also on gas consumption into the building. The current deployment consists of a number of

wireless sensor motes retrofitted throughout a residential building converted for office use. The WSN nodes

are based on the Tyndall modular mote platform running the Contiki operating system and communicating

with a mesh network running IPV6 through 6LoWPAN over IEEE 802.15.4 at 2.4GHz. Each node is

configured for a specific task within the framework of enabling energy efficiency and these tasks can be

broadly described as, environmental sensing, metering (gas and electricity) and actuation. The motes are

controlled through the LinkSmart middleware platform which is an open source hardware agnostic system

for building energy management, which hides the underlying physical layer allowing ease of development

for web based applications, which is also demonstrated as part of this work.

1 INTRODUCTION

Energy demand in residential and small commercial

buildings is increasing dramatically and currently

accounts for 65% of energy consumed on a national

grid. European Union directives are targeting

dramatic improvements in Energy efficiency of at

least 20%, combined with a reduction of green house

gases by 20% for the year 2020 (Directive of

European Parliament, 2004/2006). Managing and

reducing the energy demands from residential and

small businesses will require a low cost and easily

implementable strategy that can be easily adopted by

non-technical users.

Building management systems (BMS) currently

are not cost effective solutions for home or light

commercial users. Cost estimates for wired systems

can range from €1.60 per metre for a new

construction up to €5.00 per metre for retrofit

applications for wired BMS installations (Nan Li,

2010). This is a significant cost and does not take

into account the cost of the BMS itself. Wireless

systems can negate wiring cost with wireless motes

ranging in costs from 20 -100 euro and the cost is

continually dropping. Outside of potential cost

savings wireless systems can also offer other

advantages over traditional BMS solutions. These

are (Nan Li, 2010; Jun Zhang et al., 2011)

 Ease of deployment

 Easily reconfigured

 Easily expanded and upgraded

 Low Maintenance

 Range of operational environments , they can

be deployed in areas where wiring may not

Moore L., Hayes M., Flynn B., Mathuna C., Frecon E., Ericksson J., Kool P., Rosengren P., Fernandez A., Rosik J. and MacSuibhne D..

Wireless Network Deployment as Low Cost Building Management System Solution.

DOI: 10.5220/0004379300640070

In Proceedings of the 2nd International Conference on Smart Grids and Green IT Systems (SMARTGREENS-2013), pages 64-70

ISBN: 978-989-8565-55-6

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

be practical

 Can be integrated into smart metering

frameworks (Kaplantis, 2012)

There are a number of commercially available

technologies primarily for home automation.

Hardware such as EnOcean, Z-wave, and KNX that

have been developed that can carry out energy

monitoring and actuation commands. Currently

these systems focus on the Home environment and

while they could be adapted as light weight BMS

solutions they are currently not optimised for such.

These solutions also tend to focus primarily on

electricity consumption of devices without offering

cost effective methods of monitoring gas

consumption (Anders, 2011). Industrial based

wireless hardware and protocols exist on the other

extreme such as WirelessHart. These systems again

are not optimised for the intermediate BMS solution

and are generally targeted towards very specific

industrial applications. Additionally since these

technologies tend towards proprietary technologies

they do not offer a complete retrofit solution (they

are not entirely future proof and if existing wireless

infrastructure exists such as a smart meter, they may

not be easily interoperable) A number of researchers

have looked at physical architectures and

deployments using Zigbee wireless networks (Yang,

2009; Jinsoo, 2009) for the areas of home

automation. These papers have focused primarily on

the design of the Zigbee mote and networks and do

not discuss in any great details deployments or

complete system requirements for a wireless BMS.

Other authors have looked at using 6Lowpan

enabled networks (Bernd and Thomas, 2011) and

have focused on the advantages 6Lowpan, such as

IPV6 compatibility and internet accessibility.

For a completely adaptable retrofit installation

that can act as a BMS alternative the system should

be broken into three distinct parts that are

completely interchangeable. These are

1. Hardware infrastructure

2. Middleware platform

3. Application

The deployment presented here has successfully

retrofitted a residence that is over one hundred years

old with a light weight BMS solution, utilizing low

cost wireless sensor nodes that are flexible and non-

invasive in their design enabling them to easily

integrate onto the existing utility framework at the

pilot site.

The deployment looked at all the components

required to create a lightweight BMS solution such

as the sensor motes, a middleware platform and web

based application that could interface through the

middleware to the deployed hardware acting as the

monitoring and intelligence of the BMS system.

The rest of this paper describes the set-up,

running and evaluation of a real world sensor

network deployment demonstrating device

interoperability and a non-invasive deployment that

provides data to a communications agnostic

middleware platform.

2 NETWORK OVERVIEW

The main objectives of the work carried out were to

 Retrofit an existing building with a wireless

sensor enabled energy monitoring and

management system

 Deploy wireless sensor motes in a real world

“living-lab” environment

 Use a “self healing” mesh network to create a

robust network infrastructure

 Utilize an IPV6 protocol for web interoperability

 Integrate with the LinkSmart middleware

platform for network management

 Report to a web based application for data

monitoring

2.1 Location

A suitable building was selected for deployment of

the wireless sensor network. The building selected

was the Crossleigh building located at University

College Cork (UCC) Figure 1. It is an old residential

building dating back to the middle of last century

that was adapted for use as offices and computer

laboratories for teaching staff and students in the

School of Applied Social Studies. The building’s

heating system is water based. There is a small

boiler house where a gas driven boiler heats the

water, which is then circulated in the building by a

set of pumps, also located in the boiler house. A gas

meter is located in a meter box in the front garden.

In this application the deployed system will control

the pump that serves the second floor, and use it to

regulate the temperature on that floor. The building

itself consists of three floors with the gas boiler

located in a separate extension on the back of the

house and the gas meter in a box out in front of the

building. This building was considered an ideal test

site as it represented both an old residential building

and small commercial enterprise.

WirelessNetworkDeploymentasLowCostBuildingManagementSystemSolution

Figure 1: Crossleigh House.

2.2 Deployment Architecture

In order to retrofit an old building such as Crossleigh

house with a light weight BMS solution three main

architectural components are required. These are

 The physical hardware

 The middleware platform

 The BMS application

The system architecture is shown below in Figure 2.

The physical hardware is the monitoring and control

infrastructure for the building and is completely

wireless. The middleware platform connects the

application for monitoring and control with the

deployed hardware. The middleware creates an

agnostic environment for the application developer

to work with, allowing applications to focus on

higher level BMS and energy solutions rather than

integrating with the lower physical layer. The

middleware can be expanded to work with a range of

protocols and devices creating a complete

interoperable framework for retrofit deployments.

Figure 2: System Architecture.

2.2.1 The Wireless Hardware

The wireless sensor platform used for this

deployment is the Tyndall 25mm platform known as

the Tyndall Mote 0. The mote design is based on a

modular layout which enables the user to integrate

any combination of sensors, communications and

power source. The flexible nature of the mote makes

it ideal for an experimental wireless sensor

deployment such as this.

The motes deployed here operate in the 2.4GHz

ISM band using a Texas Instruments CC2420 as the

transceiver. They are controlled by an

MSP430F5437 microcontroller and their peripheral

layer includes: - Temperature, Humidity, Light, RS-

485 interface, UART interface, Modbus, SCADA,

KNX, and DALI

The motes in this project all run the Contiki OS

created by the Swedish Institute for Computer

Science (SICS) (Dunkels et al., 2004).The node

application layer sits on top of the Contiki operating

system, and is capable of executing small C

programs, such as programs for communicating over

modbus to an electric meter or pulse counting on a

gas meter. The Tyndall node is shown below in

Figure 3 and the architecture in figure 4.

Figure 3: Hardware platform used.

Figure 4: Wireless Hardware Architecture.

2.2.2 The Middleware

One of the key components of the deployment is to

use an energy orientated middleware platform

making it possible to network heterogeneous

physical devices into a service-oriented architecture.

The aim of the middleware is to hide the complexity

of the underlying device and communications

SMARTGREENS2013-2ndInternationalConferenceonSmartGridsandGreenITSystems

technologies, thus making it easier to develop

applications. Developers can focus more on high

rather than low level issues and functionality. This

deployment utilizes the LinkSmart (formally Hydra)

service orientated middleware software to achieve

these aims (Eisenhaur et al., 2009). This middleware

platform was developed by CNET, Fraunhofer

Institute of Technology (FIT) and Telefonica and is

an open source platform that can be adapted to any

application domain (in this case energy efficiency).

A context manager (Frécon, 2012) was also part

of the middleware platform which focuses

specifically on energy awareness. The middleware

platform communicates with a deployed wireless

network over an RS232 connection to a PC. A

border-router, communicates with the middleware

via a SLIP protocol and acts as the interface between

middleware and hardware. The middleware offers an

API to higher level developers. A device that is

incorporated through the LinkSmart platform is

presented to the developer as software abstraction

within the LinkSmart network, commands can be

sent to the network and data taken from the devices

without needing to know any of the underlying

physical process. This device can then be made

accessible via web services to the application layer.

2.2.3 The Application Layer

As mentioned one of the main advantages of using

the LinkSmart middleware platform is it hides the

complex physical layer from the higher level

application developer. This enables shorter

development times in creating custom applications

for building monitoring. These applications can be

interfaced with any hardware deployed that makes

up the infrastructure of a light weight BMS as the

middleware creates complete platform agnostic

layer. Since the middleware presents physical

devices services to the application layer as web

services the applications can be entirely web based

and accessible from any internet enabled device.

Figure 5 below shows a sample of a developed

Figure 5: Graphical User Interface for Crossleigh house.

business GUI which was created specifically for

Crossleigh house by the ME3gas partner

ResourceKraft.

While the above GUI was developed specifically

for Crossleigh house the web based nature of the

middleware enables easy integration to open source

applications again reducing costs if required. For

example the data for this application has also been

integrated with the online service COSM which is a

free online sensor data monitoring service. A

screenshot of it is illustrated in Figure 6 showing

electricity data for the ground floor of Crossleigh

house.

Figure 6: COSM Screenshot.

The first step in retrofitting this building with an

intermediate BMS solution was to deploy the

wireless hardware that would act as the sensing and

control infrastructure for the building. Since wireless

technology was chosen for the physical

communications a radio survey was first carried out

of the site in order to determine the effective range

of individual wireless motes as well as to determine

the number and placement of each individual mote

in the network. The radio survey was a simple send

receive test, where one node was configured to send

a packet of data and another node configured to

receive and display the data packet on a terminal

program through a UART port. The motes were

placed as shown in figure 7 below based on the radio

survey carried out.

Figure 7: Placed wireless sensor motes.

WirelessNetworkDeploymentasLowCostBuildingManagementSystemSolution

The Motes are segmented by task into three main

categories. These are 1. Metering and sensing 2.

Routing and 3.Actuation The metering part of this

application is split into three areas which are

Electricity, Gas and Heat Metering. Two electricity

meter motes were installed, one mote for the ground

floor of the building (M13) and the other for the 2

and 3

floors (M10) A mote was connected to the

pulse output of the gas meter in Crossleigh at M9

(Figure 8) and four meters were connected to pulse

output heat meters M2-M5 Each heat meter

corresponds to one heating zone within Crossleigh

house. The four zones are (i) Ground floor, (ii)

Computer lab (attached to ground floor) , (iii) Floor

2 and (iv) Floor 3.

Figure 8: Gas meter mote.

Based on the radio survey carried out at the start

of this work two motes were placed purely for

routing purposes at M8 and M7.

Actuation is carried out via M6 which controls

the heating for the third floor of the building. M11 is

an under sink heater and this also controlled by the

network.

2.2.4 Network Topology

The wireless network is mesh configured. It operates

UDP for sending data and a TCP-IP for

configuration with an IPV6 header over 802.15.4

utilizing 6LOWPAN for the sensor network itself. It

was envisioned that the network would run on only

TCP. One advantage of using TCP was that any

“normal” tools can be used to monitor and debug the

network and this was proven to work very well for

configuration with a REST Client plug-in for

Firefox. Data is transferred through the motes using

JSON expressions. Problems occurred when trying

to deliver data packets over TCP. Since another

advantage of TCP is the reliability that is offered

from a handshaking protocol, call-backs occur which

raises the network overhead and could cause the

network to “freeze” resulting in situations where no

data was able to get through. As a result TCP was

used for configuration and actuation commands

while UDP was used for sending data packets. This

worked to improve the network stability as non-

critical data packets could be sent via UDP (which

were the majority of data packets). Critical data

packets such as actuation commands were still sent

offer TCP to guarantee delivery.

3 RESULTS

In order to justify the cost of any building

management system savings on utility costs need to

be demonstrated. While actuation was implemented

in this deployment, due to contractual agreements

with the users of the building, no advanced control

strategies were implemented in order to avoid

impact to the quality of their utility services. All

actuation carried out was in-line with existing

building constraints. So for example the actuator

controlling the under sink heater the original system

switched on the heating at 7:00 and off at 18:00

and?? this system did the same. Despite these

restraints based on the monitoring alone and the

integrated application it was possible to show where

further cost savings could be made. Using the

deployed system it was possible to monitor the

electrical consumption of the building broken into

two zones (a) Ground floor which also houses a

computer lab used by students and (b) combined

second and third floors). The electrical consumption

is shown in Figure 9. This graph shows the

consumption per zone as well as the total peak

consumptions. Figure 10 shows the total amount of

gas being consumed in Crossleigh house.

This data is presented as kWhrs for electricity

and gas. For most people the main consideration is

cost. Through the business GUI this data can easily

be converted to monetary value if the pricing is

known. This allows the user to quickly check

various plans and costings on offer from utility

companies and determine the savings that can be

made directly to them. For example, based on the

data available and looking at various plans available

within Ireland (Residential price plans used,

company names not used as this is only a snapshot

of what’s available and not indicative of the whole

market) where this deployment is situated,

comparisons can be made as shown in Table1 for the

week.

Total Gas Consumption = 1750kWhr

Total Electricity Consumption = 700kWhr

SMARTGREENS2013-2ndInternationalConferenceonSmartGridsandGreenITSystems

Table 1: Prices applied to gathered data.

Gas

(c/kWhr)

Electricity

(c/kWhr)

Total

Gas €

Total

Electric €

Utility 1 5.540 17.161 94.5 120

Utility 2 5.628 16.93 98.49 118.51

Utility 3 5.894 17.93 103.145 125.51

The data in Table 1shows how using this data

can provide estimates of where to make initial

savings and this is before advanced control

techniques are employed.

As can be seen from the table above switching

providers based on the monitored data and real

values of up to 9% on gas bills and 5% on electricity

can be made. For the week shown this works to an

actual saving of nearly €15. This a very basic

analysis based on basic price plans and one weeks

data but the potential for ensuring cost savings and

return on investment has been demonstrated.

There is only one gas meter within Crossleigh

but the gas is only used to feed the heating supply

within the building. The heating within Crossleigh is

split into 4 separate zones. These are the ground

floor, the student’s computer laboratory, the second

floor and the third floor. As already mentioned there

are heat meters attached to the heating pipe outlets

for each zone and each meter has a wireless mote

attached to it. Thus a breakdown of gas consumption

in Crossleigh could be derived and is shown in

Figure 11. The largest area is the ground floor and

not surprisingly this consumed the most in terms of

gas. The second largest area is the second floor and

this was not surprisingly the second largest

consumer of gas. The computer lab was third largest

area but this consumed the least gas due to the fact

that heat generating equipment is located here and

less heating was needed. From this breakdown it is

easy to see what each area of the buildings is

consuming in terms of heating.

Figure 9: Electricity Consumption 1 week.

Figure 10: Gas Consumption of Crossleigh 1 week.

Figure 11: Gas use per zone in Crossleigh.

4 CONCLUSIONS

Presented here is a real world application of a

wireless network for energy metering and

management that has successfully been retrofitted

into an existing 100+ year old building. This

deployment demonstrated a clear alternative for

building management that bypasses the more

expensive & invasive building energy management

system with

 a deployment of all components that would be

required for a wireless BMS deployment

 A 6lowpan network for wireless sensing and

actuation in a real world environment

 A middleware platform to act as an

intermediary between application layer &

physical layer

 A business application for monitoring of the

deployed network

The installation did not require the addition of any

capital equipment such as new boilers or gas meters

and for the most part existing equipment already in

the building was adapted with “add-on” wireless

sensor motes in the form of the Tyndall mote.

Trouble free wireless mote deployment in this

scenario was resultant from the radio survey of the

site carried out early on in the development.

Although relatively simple it still proved a valuable

and effective tool in deploying the sensors in

WirelessNetworkDeploymentasLowCostBuildingManagementSystemSolution

suitable locations. This helped in avoiding problems

as the network was expanded. Another observation

related to the issues with TCP mentioned above.

HTTP over IPV6 is believed by many to be the

future of wireless sensor networks due to the ability

to access sensors directly over the web. This

deployment has shown that there can be issues with

this method of communications, future work on this

project is investigating the use of other strategies

such as the use of websockets reduce header

overhead when transmitting sensing and actuation

data over TCP. Other future work is to deploy this

type of set-up on a larger scale in countries such as

Sweden & Spain.

REFERENCES

Directive of the European Parliament & of the Council on

energy & efficiency & amending and subsequently

repealing directives 2004/8/EC and 2006/32/EC.

Jun Zhang, Guangming Song, Hui Wang, Tianhua Meng.,

2011 “Design of a wireless sensor network based

monitoring system for Home Automation” IEEE

International Conference on Future Computer

Sciences and Application.

Yang Li1 , Ji Maorong, Gao Zhenru,Zhang Weiping,Guo

Tao, 2009 “Design of a home automation system

based Zigbee wireless network” The 1st International

Conference on Information Science and Engineering.

Nan Li, “Exploring the use of wireless technology in

building energy management” Proc International

conference in Civil & Building Engineering, 2010.

Jinsoo Han, Haeryong Lee, Kwang-Roh Park, 2009

Remote-Controllable and Energy-Saving Room

Architecture based on ZigBee Communication” IEEE

Transactions on Consumer Electronics.

Bernd Michael Dörge, Thomas Scheffler, 2011 “Using

IPV6 & 6Lowpan for home automation networks”

IEEE International Conf on Consumer Electronics.

M Eisenhaur, P. Rosengran, P. Antolin 2009. “A

Development Platform for Integrating Wireless

Devices and Sensors into Ambient Intelligence

Systems” IEEE 6

International Communications

Society Conference on Sensor Mesh and Ad-Hoc

Communications Networks.

Karsten Menzel, Dirk Pesch, Brendan O Flynn, Marcus

Keane, Cian O’ Mathuna, 2009“Towards a Wireless

Sensor Platform for Energy Efficient Building

Operation”.

Kaplantis S., 2012 “Security and Smart Metering” 18

European wireless conference.

Anders, A. "EnOcean Technology --- Energy Harvesting

Wireless," EnOcean White paper 2011.

E. Frécon, Dec. 2012 "Bringing Context to the Internet of

Things", In Proceedings of the 19

Tcl Conference.

A. Dunkels, B. Grönvall, and T. Voigt. Nov. 2004.

“Contiki – a lightweight & flexible operating system

for tiny networked sensors. In Proceedings of EmNets.

SMARTGREENS2013-2ndInternationalConferenceonSmartGridsandGreenITSystems