BLOG

ZigBee drawbacks in a commercial environment
Overview
We are often asked why we don’t use ZigBee for our Building Automation System. The reason is simple – ZigBee does not scale well and in our view is unreliable in the commercial building environment because it competes with WiFi for radio transmission time.
ZigBee was designed for homes – not commercial buildings:
“The ZigBee® Alliance, a global ecosystem of companies creating wireless solutions for use in consumer, commercial building and utility applications, today announced the completion and immediate free public availability of the ZigBee Home Automation (HA) public application profile that offers manufacturers a standards-based approach to introducing new wireless home automation products globally, eliminating the need for proprietary technology. HA allows reliable and interoperable home automation applications to be developed by product manufacturers for consumers everywhere..“The ZigBee Home Automation Profile provides manufacturers the universal standard application they have been demanding in order to launch a thriving global home automation market allowing consumers everywhere an easy way to reap the numerous savings found in a standards-based wireless home automation network,” said Bob Heile, chairman of the ZigBee Alliance.”

We emphasized Home Automation in the ZigBee Alliance press release to highlight the fact that ZigBee was originally designed and intended for a home environment – not for commercial building like a hotel. Those are vastly different.

In our opinion, ZigBee has huge flaws for use in commercial buildings because of limitations inherent in the ZigBee architecture that was intended for simple home automation instead of large scale, dense, complex commercial building automation.

Guest dissatisfaction with WiFi in hotels is well known and hoteliers are moving to improve the power of their WiFi networks in response to guest demands. This problem will worsen as hospitality properties revamp and improve their WiFi networks to increase the speed and coverage. This will have a large detrimental effect on ZigBee.

This white paper will lay out our reasons for avoiding ZigBee by using independent third party (usually university research) to support our reasons.
The radio environment ZigBee operates in
ZigBee operates in a frequency band set aside by the FCC (in the US) that is known as the ISM bands for (Industrial, Scientific and Medical). One of the attractive things about the ISM band is that there is no FCC license to pay for like is the case for a TV station or a cell phone carrier and there is no FCC paperwork to fill out.The FCC has coordinated with like agencies in other parts of the world and the 2.4 GHz spectrum is also license free world wide.

ISM is actually several different bands at several different frequency ranges, but for this paper we only concern ourselves with the 2.4Ghz portion of the ISM frequencies.

This world wide frequency coordination then encouraged hardware manufacturers to make devices that can be used world wide. The most common of these being WiFi, with Bluetooth a close second. Travelers around the world can use the same WiFi device in virtually any country anywhere on the globe.

This same ability appeals to the makers of sensors. One frequency world wide means only one set of hardware. So ZigBee in the 2.4 Ghz ISM band is license free and can be used everywhere WiFi can be used. Bluetooth is also in this band.
The 2.4 GHz radio spectrum and its uses
There are several technologies that use the license free 2.4 GHz radio spectrum. They include WiFi, ZigBee, Bluetooth, car alarm systems, wireless video cameras, wireless video game controllers and others.

For the purposes of this white paper we will only address the interplay between WiFi and ZigBee and focus on hotels which have the requirement (driven by guest demand) to provide pervasive, high quality, high speed Internet via WiFi. WiFi is no longer a option for hoteliers – it has become an expectation of all guests.

Here is the spectrum available in the 2.4 GHz band and the original mapping of the WiFi (802.11) and ZigBee channels (802.15.4). This diagram assumes that the WiFi networks are only using WiFi channels 1,3 & 6 – leaving a few bits of radio spectrum where ZigBee MIGHT expect less interference from WiFi.

However in a hospitality environment it is highly unlikely that only 3 WiFi channels will be used, because each WiFi channel maxes out at 300 Mbps which must be shared across all users on that channel. It is more likely that a hotel will make use of all 13 available WiFi channels in order to provide the maximum WiFi service to their guests. Network designers will place WiFi Access Points throughout the property and will plan geographic separation of adjacent channels to minimize the co-channel interference among the WiFi Access Points.

Problem Summary in layman’s terms
WiFi is everywhere in a hotel and it is on all the time and being used constantly and WiFi has tremendously more transmit power than battery powered ZigBee devices put out. Because WiFi and ZigBee must share the radio spectrum and do not coordinate – WiFi vastly overpowers and dramatically degrades ZigBee transmissions.

The Red line is ZigBee with no WiFi present. Blue line shows one channel of WiFi (channel 6) actually affects the ENTIRE ZigBee spectrum, with the most severe problem in channels ZigBee 16 and nearby.


WiFi characteristics overview
Architecture is user devices connecting to one of many Access Points
Access Points use grid power, often WiFi users do as well
Max power output is 4 watts (in the US)
Very high speed – up to 300 Mbps / channel
High periods of activity (streaming video, file downloads, etc)
Wide channels – up to 40 Mhz of radio spectrum for for N channel
TCP/IP protocol (the protocol of the Internet)
Typically managed by IT department
ZigBee characteristics overview
Architecture is sensors talking to routers to a single coordinator
Sensors battery powered therefore use low transmit power to conserve battery
Very low powered radio (up to 10 times less output power than WiFi)
Very low speed – about 200 kbps / channel
Low periods of activity – generally only on sensor activation
Narrow 5 Mhz radio channels
Proprietary protocol not easily integrated into corporate intranets of the Internet
Typically managed by building management / engineering department
The problem expanded
There are two major problems with ZigBee. They are reliability and battery life.
Reliability
Not only does WiFi overpower ZigBee transmissions, but ZigBee transmissions can interfere with other ZigBee transmissions. This would be especially true in a situation where multiple ZigBee vendor’s systems are active and not coordinated from a frequency standpoint.

“Wu et al. [34] have shown that the number of orthogonal IEEE 802.15.4 channels
in the 2.4 GHz ISM band is only eight, despite the actual number of channels with
5MHz spacing available is sixteen. The authors carry out experiments using MicaZ nodes (that employ the CC2420 radio chip) and show that transmissions in adjacent channels decrease the packet reception rate. The interference generated in the adjacent channel can decrease the packet reception rate as much as 50%.”

The packet loss rate of a wireless sensor network operating in the presence of Bluetooth interference is often between 3% (as reported by Bertocco et al. [22] and Penna et al. [23]) and 5% (as reported by Huo et al. [24]) up to a maximum of 9–10% (as shown in the experimental results of Boano et al. [18] and Sikora and Groza [20]).

The coexistence between IEEE 802.11b/g/n and IEEE 802.15.4 devices represents a challenge for several reasons. Firstly, Wi-Fi devices are nowadays ubiquitous, especially in residential and office buildings where many Access Points (AP)are installed. Secondly, IEEE 802.11 devices operate at significantly higher power(≈24 dBm) than traditional low-power sensor nodes. Thirdly, Wi-Fi channels have bandwidth of 22MHz and can therefore interfere with multiple IEEE 802.15.4 channels at the same time. Fourthly, IEEE 802.11 supports high-throughput transmissions that generate interference patterns that are difficult to predict, as they depend on factors such as the number of active users, their activities, the protocols they use (UDP or TCP), or the traffic conditions in the backbone.

Several works in the literature investigate the impact of IEEE 802.11 communications on the reliability of IEEE 802.15.4 transmissions, and show that wireless sensor networks suffer from high packet loss rates in the presence of Wi-Fi interference [19, 20, 22, 23, 25, 26].

Under certain conditions, also IEEE 802.11 devices can suffer from the interference of nearby wireless sensor networks. The communications
between sensor nodes can indeed trigger a nearby Wi-Fi transmitter to back off.[8, 26].

The actual packet loss rate that IEEE 802.15.4 networks experience in the presence of IEEE 802.11 transmissions depends on the Wi-Fi activity, as well as the location of the nodes. Boano et al. [18] have varied the Wi-Fi traffic pattern, showing that activities such as continuous radio streaming are not too critical for sensor net communications, as it results in approximately 15% packet loss rate. On the contrary, activities such as video streaming (≈30% packet loss rate) and file transfers (≈90% packet loss rate) can destroy the majority of wireless sensor networks transmissions and cause long delays, drastically decreasing the performance of the network.

The characteristics of the interference patterns emitted by domestic microwave
ovens depend on the model; nevertheless all the ovens present the same basic properties. Firstly, with respect to the frequency spectrum, microwave ovens can potentially interfere on a large portion of IEEE 802.15.4 channels

During the active period, the communications of low-power sensor nodes in proximity of a microwave oven are likely destroyed (because microwave ovens operate at up to ≈60 dBm),

IEEE 802.15.4 channels in the 2.4 GHz ISM band are not orthogonal to each other, and hence wireless sensor networks operating on adjacent channels may interfere with each other [31–36].

Wu et al. [34] have shown that the number of orthogonal IEEE 802.15.4 channels
in the 2.4 GHz ISM band is only eight, despite the actual number of channels with
5MHz spacing available is sixteen. The authors carry out experiments using MicaZ nodes (that employ the CC2420 radio chip) and show that transmissions in adjacent channels decrease the packet reception rate. The interference generated in the adjacent channel can decrease the packet reception rate as much as 50%

These results were confirmed by the experiments by Ahmed et al. [31]: the packet reception rate decreases significantly in the presence of activities in adjacent IEEE802.15.4 channels.

This technique assumes that, in addition to Wi-Fi, no other interference source will ever interfere in that specific channel with the communications of the wireless sensor nodes, and hence it is highly unreliable. In more complex scenarios, involving dense wireless sensor networks covering large areas, different interference sources may be present throughout the network, and the quality of channels may differ from node to node.

Another problem comes from the fact that real-world interference cannot be easily
repeated. Gnawali et al. have highlighted that evaluating protocols by running them one at a time on real-hardware is not optimal since no experiment is absolutely repeatable [105], and this applies especially to experiments involving wireless systems, as wireless propagation depends on a myriad of factors. For example, experiments exploiting “ambient interference” surrounding the wireless sensor net?work test bed may not be a suitable option to compare several protocols or applications under realistic interference, as the interference patterns are not fully controllable and cannot be recreated precisely.