A closer look at wireless-based communication protocols reveals a variety of solutions: For example, wireless M-Bus, KNX-RF, Enocean (with its own protocol) and Z-Wave, all of which work in the 868 MHz frequency range. However, these are different protocols that are incompatible with each other - i.e., they are terminal devices from different manufacturers and cannot communicate with each other. One thing they all have in common is the need for fee-based membership in the corresponding organization. If you opt for one of the standards given above, the devices are of course compatible.
In addition to these quasi standards, proprietary solutions from the different semiconductor manufacturers are also available. Although these are free of charge, they are only interoperable for the devices using the corresponding protocol.
This includes SimpliciTI from Texas Instruments, for example. This is a stack that supports star, point to point and cluster tree network topologies with one router.
Particularly in Europe, customers of Texas Instruments have a lot of experience in RF development in the 868 MHz band and are somewhat skeptical of the 2.4GHz band. Both frequency ranges are ISM (Industrial Scientific Medical) bands. The restricted transmission range and the “overloading” of the 2.4GHz band, which hosts wireless LAN, Bluetooth and microwave transmissions, are regarded rather critically by developers.
Nonetheless, this band is of great interest since it has global approval in contrast with the 868 MHz and 433MHz bands (in Europe) and the 915MHz band (in the USA). The ZigBee protocol is available for the 2.4GHz band for the profiles of Home Automation, Smart Energy and RF4CE for remote controls in order to ensure global interoperability. The ZigBee specification, of course, also supports the protocol in an 868MHz transceiver where it is implemented on a microcontroller or SOC (System on Chip - i.e., both the microcontroller and transceiver are integrated onto the chip), but the selection of components is somewhat restricted in this case. Companies love competition and there is certainly plenty of that when one considers 2.4GHz transceivers and SOC chips. Membership of the ZigBee alliance is also subject to a charge. In the most basic case of “adopter” membership, there is a one-time charge of $ 3,500 in the year in which the product is launched on the market.

In order to remove any doubts among the 868 MHz community regarding range and noise susceptibility, a ZigBee network with star-shaped topology was installed in a house. The major advantage of an interconnected network - i.e., the automatic transfer of information via other nodes in the network from the source (e.g., a temperature sensor) to the destination (e.g., a fan), is not used here because the routers (as the nodes for forwarding the information are referred to) require a constant power supply from the mains network.

Test description
Coordinator: SmartRF05EB plus CC2530EB (see photo).
End devices: Battery board SOC_BB 1.1, power supplied by 2 AA batteries (see photo)
Transmission power of the sensors: 4.5dBm
Timing: Every 10 seconds the sensors transmit the temperature measured above the corresponding radiator to the coordinator. The data could be processed there. The data could also be forwarded via ZigBee, GSM or over the Internet.

Description of the townhouse
Steel-concrete ceilings with a thickness of between 31cm and 33cm
Wall thickness: 13cm
Length of house: 11m
Width of house: 6m
Maximum number of ceilings between two nodes: 2. All sensors are positioned on the radiators (see photo).
Since WLAN is also installed in the house, a free channel had to be found in the 2.4 GHz band. A scanner from Metageek was used in combination with the Chanalyzer Lite software for this purpose. The results are shown below. It can be seen that there is a gap in the spectrum at 2420MHz. The entire network was therefore set to the ZigBee channel 14. A similar function could also have been implemented in the software of the protocol stack so that the network automatically searches for a free channel. In this context, we should also briefly address the topic of frequency agility. Using this function, which is integrated into the ZigBeePRO stack, the entire network can switch to another channel if the channel originally selected suddenly experiences interference.
The entire wireless communication was recorded using the sensor network analyzer (SNA) from Daintree. Along with the information in each individual node, the SW also analyzes the network topology. As shown in Figure 5 below, the coordinator (with mains power) is located on the ground floor, while the sensors are distributed throughout the house.
Another important benefit of this SW is its ability to measure the strength of each signal. In the illustrated test setup the SNA sensor was placed directly beside the coordinator board. From Figure 6 it can be seen that the transmission power of the coordinator with the address 0x0000 was measured to be 4dBm. The RF power of all nodes, including that of the coordinator, was set to 4.5dBm. This shows that the measurements for the other nodes are in all regards representative. The weakest signal was measured to be -70dBm by the sensor on the 3rd floor (i.e., 2 floors above the ground level). This means that in reference to the IEEE 802.15.4 specification, which specifies a sensitivity value of 85dBm, there is a margin of 15dB.
The integrated CC2530 system-on-chip solution (microcontroller together with transceiver) from TI is specified with a sensitivity of typically -97dBm for an error rate of 1%.
As a result, there should be no problems regarding transmission range.
The recorded data shown in Figure 6 indicate that the nodes are transmitting every 10 seconds and that the data is transmitted in encrypted form.
Battery lifetime is a deciding factor when it comes to sensors. Customers typically expect sensor applications to last up to 12 years and more.
This calls for batteries with power ratings of up to about 3Ah - which can be provided by two AA cells, for example.
In the practical test examined here, the battery lifetime is to be calculated by first obtaining the integral of the current over the time. The current was indirectly measured - i.e., the voltage drop across a 1W resistor connected in series with the power line is measured. In the diagram (Figure 7) we therefore have 10mV = 10mA. The diagram also shows that the sensor node requires only about 1µA in low power mode 2 (the realtime clock and timer are activated) before and after sending data.

Description of the current profile
Along with the transmission current, the current consumed in sleep mode and the loss of power due to self-discharge (a figure of 0.2% per month is assumed for this) must also be considered.
a) The energy consumed (charge volume) for one-time transmission of data at a level of 95.10 mAms corresponds to 95.10 mAms/1000/3600 mAh = 26.4•10-6 mAh. If the data is transmitted every 10 seconds, the current consumption is = 6.84 mAh/month
b) The charge consumed per month in sleep mode is calculated as 1 µA•30•24h = 0.72 mAh/month
c) The charge loss is calculated to be 3000 mAh•0.2% = 4.8 mAh/month

This therefore yields a total charge consumption in the example of (6.84 + 0.72 + 4.8) mAh/month = 12.36 mAh/month.
If we assume that an 80% share of the battery is used for the transmitter and 20% for other functions (e.g., LCD and/or LED indication), the battery lifetime is calculated to be 0.8•3000 mAh/12.36 mAh/month = 194 months = 194/12 years = 16 years.

Summary
Two basic points were illustrated in this article:
1) It is possible to use a ZigBee network in a star-shaped topology in a townhouse. It then makes sense to install battery-operated sensor nodes.
2) The battery lifetime of 11 years is a thoroughly acceptable value.

Sources
Calculator_version_0.9.xls
( www.learnzigbee.com/calculators)

Brief biography
Hans-Günter Kremser works as an Analog Field Application Engineer at Texas Instruments in Munich. After completing his studies in information technology in Cologne, he worked at EADS in Ulm as a development engineer and then at two semiconductor manufacturers in Munich before joining TI in June, 2006.

For more information on TI wireless,
Contact: Irina Marin ( irina.marin@ecas.ro )