4 4 ZigBee and Bluetooth are differentby design and are optimised for

different applications. Realindustrial wireless networks will

inevitably be hybrids includingboth in complementary roles.

IEE Computing i Control Engineering | April/May 2005

ZigBee and Bluetooth

ONLY IN THE LAST 10 YEARS OR SO, WITH CONTINUING ADVANCES INSEMICONDUaOR, RADIO AND BATTERY TECHNOLOGY HAS SIGNIFICANT EFFORT BEENMADE TO DEFINE AND DEVELOP WIRELESS TECHNIQUES FOR DATA NETWORKS.

Bluetoothstrengths and Weaknessesfor Industrial applicationsBy Nick Baker

Most industry analysts are fore-casting explosive growth in the use ofwireless data network technologies inindustrial applications in the next fewyears.Figure 1 depicts the wireless spectrum

in terms of two key performance characteristics - wirelessradio range aiid data transmission rate. Other performancecharacteristics will be discussed later but in terms of thesetwo parameters it is important to recognise that the twoIEEE standards that underpin ZigBee (802.15.4) andBluetooth (802.15.1) are intended to differentiate them fromeach other.

The IEEE defmes only the Physical (PHY)and Medium Access Control (MAC) layers inits standards. Eor both ZigBee and Bluetoothseparate alliances of companies worked todevelop specifications covering the network/link, security and application profOe layers sothat the coniniercial potential of the standardscould be realised.

Bluetooth originated in 1994 when Ericssonbegan to examine alternatives to cahles tbatlinked mobile phone accessories. In 1998Ericsson was joined hy IBM, Nokia. Intel, andToshiba to form the Bluetooth Special InterestGroup (SIG) which defined the initial

specification. In mid-1999 the SIG approached the IEEE andasked them to formally adopt the Bluetooth specifications.The 802.15.1 Standard was published in 2002. Thousands ofPromoter. Associate, and Adopter companies have sincejoined the SIG which develops, publishes and promotesBluetooth and runs a qualification program to foster deviceinteroperability

ZigBee's origins date only from 1998 when Motorolastarted work on this type of low power mesh networking.The IEEE 802.15.4 standard was based on Motorola's mid-2001 proposal and was ratified in May 2003. Phillips,Motorola, Invensys, Honeywell, and Mitsubishi -^

WAN

Fig I: The wireless landscape

IEE Computing & Control Engineering | April/May 2005

joined together and formed the ZigBee Alliance in mid-2002to develop and promote this technology and leverage thestandard. Ember. Freescale and Samsung joined aspromoters later. They worked together on defining thenetwork, security and application layers of the ZigBeespecification, which was ratified in December 2004. Thereare now well over 100 affiliate members of the ZigBeeAlliance representing semiconductor manufacturers,technology development companies, OEMs, end usercoinpanies and systems integrators.

TECHNOLOGY OBJEaiVESSo wbat are the objectives behind the two technologies?Looking in a little more detail we can see some cleardifferences and some similarities.

Firstly looking at Bluetooth, the SIG mission statementdefines an objective targeting short range and mobile

Direct Connection

• Wire replacement

• Point to point

Star• One central routing and controi point

> Singie-iiop-point to multi-point

' All data flows through centrai point

< Exampies are UVifl. Bluetooth. CSM

Mesh

• Multipie data paths• Muilt-hop

• Seif configuring, seif healing

• Examples are ZigBee, EmberNet SensiNet

Fig 2: Wireless network topologies

products and this is echoed by the IEEE in defining a'Personal Operating Space" (POS) of 10m radius andallowing for mobility The use of the word 'personal' linksthis technology at its core to provision of ad-hocconnections hetween devices used by humans.

The types of device interoperability profiles that bavebeen developed for Bluetooth [cordless telephony headset,LAN access, fax, printing, hands-free, etc.] and the types ofapplication areas in which products bave been developedare very much in line with the intent of the SIG and tbestandard.

Looking at ZigBee the key additions or differences interms of the alliance mission statement are low power,networked |as opposed to connected], and open standard.The 802.15.4 standard also speaks of a POS and 10m rangebut recognises the possibility for greater range at lowerdata rates.

These mesh self-bealing networks, which allow mobility

of end nodes within tbe network and, by virtue of theirmulti-hop capability can cover large areas, will have a verywide range of applications from industrial sensing andcontrol to huilding automation and security, homeautomation and even in interactive toys.

DESIRABLE CHARAaERiSTiCS OF NETWORKSLet's focus our attention on industrial wireless datanetworks and their desirable characteristics.

Range: At least 50m in "cluttered" industrial RFenvironments where there is often a lot of metal inequipment and building structure and increasing amountsof radio interference.

Data rate: In industrial sensing and control applicationsrequired data rates vary widely by application but are oftenlow and/or intermittent.

Network latency [or how long the data takes to get fromorigin to destination/: This varieswidely by application. It shouldideally be possible to tune tbenetwork availability or response totbe application i-equirement at theend-node to optimise performance.A second consideration is howlong new devices take to join thenetwork.

Power profile: Ideally under allcircumstances devices would bebattery-operated to avoid bothpower and data wiring costs andincrease tlexihility

Security: At the lowest level:How sure can I be that the data didget from origin to end pointaccurately and completely? This is

critical in sensing and control applications where humansdo not normally validate data at the operating timeinterval. At the highest level: How sure can I be that mynetwork and its data cannot be 'hacked' and the datamisappropriated or meddled with? Am I able to controlwhich devices join my network?

Operating Frequency: The main requirement here isoperation in one of the unlicensed bands for operating costand regulatory reasons. Globally tbe 2.4GHz ISM band isemerging as the preferred band, which brings increasedrisk of interference from otber devices. Much of theoverhead of wireless network protocols involves strategiesto avoid interference degradation of network integrity

Engineering and design complexity: More complexitywill drive up cost for product developers, implementers andend-users. The technology should ideally be relativelysimple to understand and engineer into products, have lowoverheads in I ^ ^ s of system operation and design, and be

lEE Computing & Control Engineering i April/May 2005

ZigBee and Bluetooth

simple to implement and support.Network topology: Increasing the number of possible

communication paths through the network increases thelikelihood that the message will be received at itsdestination, even if after multiple hops. Tbis makesnetwork traffic more complex but will increase the networkresilience and reliability Ideally the fuH range of topologies[Figure 2] sbould be supported.

Number of devices: The numher of requiredmeasurement points is increasing significantly often in a'retrofit' manner, to more completely measure industrialenvironments and processes for better control and forcompliance and audit purposes. In most real industrialapplications many tens, hundi'eds and possibly thousandsof devices could he required in a network.

Scalability/Extendability: Industrial environmentsconstantly change - growing or shrinking in size and thenumber and nature of measurement points. Sometimesthis is short term - for example. Intensifying measurementduring commissioning stages of a new plant. Tbe wirelessnetwork must be capable of accommodating these changeswithout significant support overhead to tbe enterprise.

Flexibility: The networking technology should be flexiblein terms of tbe uses to which it might be put. It should beagnostic to the type of sensors or output devices attachedand able to be implemented for different device typeswithout a lot of device-specific requirements within thenetwork or tbe protocol stack.

Resilience and reliability: It must have the real world"

Bluetooth is 'always on' and mustbe recharged frequently; ZigBee'sleeps' most of the time and hasyears of battery life , ,

performance capability to deal with transient interferenceand obstacles. It must be able to manage and adjust thenetwork configuration, ideally automatically and know orbe alerted when the network encounters a situation that itcannot resolve. During the network implementation itshould be possible to design out unnecessary single pointsof failure.

COMPARING ZICBEE WITH BLUETOOTHWe turn now to a comparison of the two technologiesin terms of tbe desirable features above and referenceFig 3.

Range: As designed and without special equipment it isclear that ZigBee has the potential to operate over a greaterrange especially in 'low clutter'-radio environments. Tbeupper range limit has really only been possible withproprietary mesh networking protocols, such as SensiNet.running over 802.15.4 radios.

Data rate: Where higher data rates are importantBluetooth clearly has the advantage and can support awider range of traffic types than ZigBee. ->

characteristic

Range

As designedSpecial kit or outdoors

Data rate „ ,,.,-,,,,„,.„

Network Latency (typical)New slave enumerationSleeping slave changing to activeActive slave channel access

Power profile

Security

Operating Frequency

Complexity

Networii Topology

Number of devices per network

Scalability/Extendability

Flexibility

Resilience and reliability

ZigBee

10-100 metresup to 400 metres

30ms15ms15ms

YearsOptimizes slave power requirements

128 bit AES and application layer user definable

868 Mhz, 902-928 MHz, 2.4 GHz ISM

Simple

Adhoc star, mesh hybrid

2 to 65,000

Very High/Yes

Very High

Very High

Bluetooth

10 metres100+ metres dep. on radio

20s3S

2ms

OaysMaximises adhoc functionality

64 bi t 128 bit

2.4 GHz ISM -^M

Complex

Adhoc piconets •^^^.

8

-:flHK^ Low/NoMedium, profile dependent

Fig 3: Comparison of desirable characteristics

IEE Compjtmg & Control Engineering | April/May 2005

Focus on remote sensing and control• Warehouses, Fle«t management Factory, Supennarhets, Office

complexes• Cas/Water/Eiectric meter, HVAC• Smoke, CO, H,0 detector• Refrigeration case or appliance• Equipment management services ft Preventative maintenance• Security services• Lighting control• Assembly line and work fiow. Inventory• Materiait processing systems (heat gas flow, cooiing, chemical)

Temp sensor

Energy, diagnostics, e-Business services

• Gateway or Field 5ervice links to sensors ft equipmentMonitored to suggest PM. product updates, status changes

• Nodes iink to PC for database storagePC Modem calls retailer. Service Provider, or Corp headquartersCorp headquarters remoteiy monitors assets, bill ing, energymanagement

Fieid Service ormobiie worker

1004 Tba Zi(BM MHanca, Inc

Materiais handDng

Fig 4: ZigBee industrial applications

ServiceProvider Retailer

Network latency: To be able to sleep for extended periodsto conserve power and achieve acceptable network latencyZigBee end devices need to wake up very quickly, transmitand/or receive and go back to sleep. The multi-hop natureof mesh networks also increases latency. Bluetooth isclearly designed for single hop device-to-device where thenodes do not sleep for much of the time and as a resultnetwork access is fast.

Power profile: Bluetooth devices are constantly alert foravailable networks for them to join. To do that they have tobe awake. The power profile is 'always on" to maximise thisad hoc networking functionality with days of battery lifeand regular recharging required. ZigBee has beendeveloped specifically to permit low power consumptionand years of battery life.

Security: Both protocols have security huilt in. 802.15.4specifies use of the 128 bit Advanced Encryption Standard

High speed Bluetooth embodiesdevice profiles for equipmentinteroperability whereas ZigBee isintended to be an open globalstandard

and the ZigBee specification defines how to handleencryption key change and multi-hop transmissionsecurity Security is also user definable within theapplication layer for ZigBee networks. Beyond encryptioneach ZigBee node retreives a unique short address from thenetwork coordinatoi' and each ZigBee network has a uniqueID. In addition ZigBee networks can also be open or lockedto new devices. Bluetooth uses 64 or 128-bit encryptionbased on the SAFER+ algorithm for authentication and keygeneration.

Operating frequency: ZigBee supports most of the widelyused unlicensed ISM bands in Europe, NA. and around theworld whereas Bluetooth operates solely on the 2.4GHzband. Although the 2.4 GHz band is becoming a de factoglobal standard (many companies in North America nowprefer it to 915 MHz) support for other bands can heimportant to industry for legacy reasons.

Complexity: We have mentioned the relative complexityof the Bluetooth protocol stack compared to ZigBee and thefact that Bluetooth embodies device profiles for equipmentinteroperability whereas ZigBee is intended to be an openglobal standard - a ZigBee compliant device from anymanufacturer should interoperate with any other.Deployment complexity and operational support of pureZigBee networks are as yet untested in the real world butmakers of proprietary 802.15.4-based mesh networkingtechnology such as Sensicast have found thatimplementation and support with networks of severalhundred nodes is relatively simple. Bluetooth complexityis, in practice, limited by the small numher of devicesallowed in each network.

iEE Computing & Control Engineering | Apnl/May 2005

ZigBee and Bluetooth

Network topology and number of devices: The increasedrange of options in terms of topology and the significantlylarger numher of devices per network would suggest thatZigBee will have much greater capahility to address thespectrum of industrial situations.

Scalabiiity/Extendabiiity: ZigBee has a significantadvantage here in terms of the ease of network growth toquite large scale implementations and the ahility to use theflexible topologies to accommodate real-world situations.

Flexibility: In theory both protocols are tlexible and cancarry any type of data. In practice the profile dependencyof Bluetooth carries some built in inflexibility. In someways this category could be seen as a qualitativeamalgamation of all the preceding categories whichsuggest that ZigBee is the more flexible approach forindustrial applications except where there is a need forhigher data rates.

Resilience and Reliability: From the purely technicalperspective ZigBee wins here in terms of the range ofindustrial situations that are likely to be encountered, dueto its data packet acknowledgement. CSMA-CA approach,encryption, mesh multi-path transmission redundancy andahility to physically worii ai'ound the buUt enviromnent dueto the hybrid network configuration options. Within itsconstraints Bluetooth is resilient - it works very well forcertain application types.

ZIGBEE APPLICATIONSFigure 4 shows the wide variety of foreseen applicationsfor ZigBee and other 802.15.4-hased proprietarytechnologies. There is a focus on remote sensing andcontrol reflecting the ZigBee mission statement, and thepossibilities are virtually limitless.

Many of these applications apparently require theadoption of ZigBee by OEMs on a large scale. Before thathappens there is a huge opportunity to retrofit enhancedsensing and control into existing huilt environments usingZigBee/802.15.4 technology through 'off-the-shelfproduction-ready mesh networking elements linked to anyof the wide range of existing sensor types and actuators.Clear advantages over classical wired installations arespeed, low cost, tlexihility and long-term re-usability all ofwhich can help increase enterprise productivity

ZigBee is not yet field-tested for these applications. Manyorganisations are developing ZigBee products but this isstill in the early stages hecause the initial specification wasonly ratified a few months ago, in December 2004. Since802.15.4 was published many companies have beendeveloping 802.15.4-based mesh networks. All the existingproducts in this sector use proprietary non-ZigBee networkprotocols on top of 802.15.4 although many are designed tosupport the ZigBee protocol stack on the same hardware.Examples are EmherNet. SensiNet and Millennial Net.

Looking at Bluetooth there is clearly an intended focuson short-range cahle replacement for medium bandwidthdevice to device connections. Beyond this. Bluetooth accesspoints can extend LANs and corporate networks toBluetooth devices.

In the industi'ial world the most likely uses for Bluetoothare for machine to machine communication and for ad hocconnectivity between mobile computing devices and fixedequipment. This could he for diagnostics, data transfer orconfiguration, especially in cases where use of temporarilyconnected cahles would be difficult such as in certain typesof hazardous environments.

Examples of current uses largely follow this trendbecause Bluetooth has been established as a useful standardfor at least two years longer than ZigBee. It has reached anearly level of maturity but is still heing promoted into newusage areas and extended in capability and refmed.

In summary it seems clear that ZigBee and Bluetoothare different by design and are optimised for different

In industry Bluetooth will most likely be usedfor machine-to-machine communication andfor adhoc connectivity between mobilecomputing devices and fixed equipment

applications. Real industrial wireless networks willinevitably be hybrids including ZigBee/802.15.4 andBluetooth in complementary roles that suit thecharacteristics of each. The key to success will be indeploying the right wireless technologies for therequirements of the application and avoiding thetemptation of trying to make one technology meet allneeds.

However, considering the wide range of typicaliTidustrial opportunities for wireless network use it seemsclear that ZigBee and 802.15.4-hased proprietary protocolscan meet a wider variety of real needs than Bluetooth. Thekey reasons are the intrinsic value to the industrialenterprise of long-term "unattended" battery operation,greater useful range, flexihility In a number of dimensionsthat were highlighted earlier and finally the inherentresilience and reliability of the mesh networkingarchitecture. •

The author, Nick Baker, is the managing director of AdaptiveWireiess Solutions. He many be reached at nbaker@adaptive-wireless.co.uk

IEE Computing i Control Engineering | April/May 2005 25