Xbee 802.15.4 Reliable Indoor Range

Hey everyone, I am working on a home sensor and automation project and have pretty much choosen to use the Xbee 802.15.4 (series 1) modules. I do not require high datarate. Most messages will be at 1 Hz or slower. The messages will only be a few bytes. Xbee seems to be a perfect solution.

I am about to make my purchase of 15 modules. My dilemma is whether or not I need the Xbee-Pro 802.15.4 modules ($32 ea) to have reliable comms through walls or if I can get by with the Xbee 802.15.4 1mw modules ($19 ea). At 15 modules, the cost difference is almost $200!

My home is a 2-story home (about 1700 sq. ft.). I plan to have a sensor/automation box in every room of my house. My network is a bit complex in that I wish to have every node know about each other. There will also be multiple control boxes. One control box will connect to a PC and the other will be a standalone unit with a keypad and LCD. The standalone control unit will be pretty much in the center of the house on the first floor and the PC box will be in a room. So basically, all sensor/automation units will have to be able to communicate with each other as well as with the control boxes. I can do this by making all units point-to-point or by using something like DigiMesh. DigiMesh allows for true Mesh where each node is an end-point and a router and a coordinator is not needed. With DigiMesh, I may be able to use the Xbee 802.15.4 1mw modules.

I decided not to use the Zigbee or Series 2/2.5 modules because I am really interested in DigiMesh. Also, I plan to power the sensor/automation boxes via battery and solar. With DigiMesh, each node can sleep. With Zigbee, the routers and coordinators should not sleep. I know there are solutions to those problems such as having a dedicated router thats always plugged in but I am pretty stubborn in that I want to use DigiMesh if I do any meshing at all.

So…I’m confused and I am hoping for some advice from someone who has done something similar. There are several options:

Should I spend $200 more to buy the Xbee-Pro 802.15.4 modules and configure it as point-to-point? This seems most reliable and simple but most expensive as well.

Should I buy the Xbee 1mw 802.15.4 modules and configure it as point-to-point and hope they can all talk to each other through walls? Would 1mw with a small 2.1 dB RPSMA omni antenna be good enough?

Should I buy the Xbee 1mw 802.15.4 modules and wait until DigiMesh comes out which should be in a few weeks? This is proably the most efficient method and my preferred option but very risky since DigiMesh is new. Also, if DigiMesh fails AND the 1mw modules can’t all talk to each other via point-to-point, then I will have wasted a good deal of money.

What do you think I should do? All comments are appreciated.

Thanks,

Steve

Your choice is good sense.

I wonder, though, if Digimesh will be here 2 years from now, as it’s proprietary and based on XBee series one. Series 2 use Ember’s chipset. Series 1 is probably much more popular now as few are using ZigBee; most use something simple atop 802.15.4.

Take a look at these options, too:

  1. use 802.15.4 without ZigBee or Digimesh. In a home automation application, each node can be configured with the “static” address of its neighbor nodes since nodes don’t move as a rule in this application. Your application can try to talk to one of 3 or so well known neighbor-nodes just by using their known MAC address in 802.15.4. Simple. Not self-forming but easy.

  2. Look at 6loPAN - public domain meshing. Not sure it has sleeping routers (ZigBee 2007 still has no sleeping routers)

  3. ISA 100 has sleeping routers in its spec. It’s a ZigBee alternative. Example implementations are starting to appear though ISA may not ratify this standard until late '08.

  4. Crossbow and Dust, Inc. have sleeping router nodes. May be too expensive though.

have you tried the XBee 1mW modules yourself in your home’s environment?

I think that choice of antenna is as important as 1mW vs 50mW (XBee versus XBee PRO). The modules can be had with:

  1. short wire antenna (2 in. or so).

  2. PC board chip antenna

  3. Coax connector for cable/antenna

I’ve used all these. The chip antennas are appealing due to size. They do have nulls in the pattern as they are not an isotropic radiator (spherical pattern). No antenna is perfectly isotropic. The wire antenna will have nulls as well. Using a coax/external antenna is ideal, but of course, higher cost, large size.

for the home automation challenge, I’d suggest using the wire or chip on most, and an external antenna on mesh nodes that are to be gateways to another network or to a PC or controller.

My estimate for 1mW radios

100-200 ft. line of sight

50-100 ft penetrating two or so walls with double-drywall. Lath/plaster or brick is a different story.

I have measured 3/4 mile line of sight for a pair of XBee PROs, one with a chip antenna and the other with a 5.6dBi antenna (coax cable version of XBee). Inside an office building with walled offices and hallways, (drywall), the XBee PRO with yielded about 200 ft penetrating perhaps 10 or so walls. Though '15.4 is 2.4GHz as is WiFi (802.11), the fact that '15.4 uses 2MHz bandwidth and OQPSK modulation makes it able to do much better with lower power than 802.11.

Wireless Engineering guidelines:

You can estimate this using a link budget and the attenuation of materials as published by NIST. But if you need every spare dB, you need to “try it”. Be sure to allow lots of margin in a link budget:

Tx power + tx antenna gain - free space path loss - wall/floor penetration loss + receive antenna gain versus receiver sensitivity. Margin is receiver sensitivity compared to received signal level (RSL). Example:

Tx power = 0dBm (1mW)

Tx ant. gain = -2dB chip antenna (wire may be +1 or +2dBi). External antenna gain can be huge, e.g. with WiFi patch antennas or whatever.

free space path loss is textbook formula for 2.4GHz and some path length (inter-node distance)

wall attenuation. I recall that NIST says about 2dB per layer of drywall at 2.4GHz. One wall = two layers. Brick/concrete, wood all differ (more).

receive antenna gain - same issues as transmitter antenna

– These sum up to RSL.

Receiver sensitivity for 802.15.4 is about -92dBm. (Some vendors claim better, say, -98 or so, but I say BS; it’s driven by the laws of physics for OQPSK at 2MHz which is 802.15.4). Sensitivity is defined as required RSL for some bit error rate, such as a 1% uncorrectable frame error rate. 802.15.4 has ACK/retransmit to correct most errors.

this will get you into the ball park. If the margin is low, you’ll want to do experiments. I’d suggest a 10dB margin to allow for fades and blocks in the path due to people moving about, and so on. A lot of link budget sins are cured by an application with prudent delayed retransmission attempts, beyond what 802.15.4 does in the MAC layer. Also message sequence numbering at the application layer to do lost packet correction.

stevech,

Thanks for the helpful response. I was hoping you would comment. Your estimate range for the 1mW radios of 50-100ft through 2 or so walls has lead to my decision to purchase the Xbee-Pro 802.15.4 modules. After reading the product manual, I have found out that the power level can be reduced to 10mW via the PL command to reduce power needed to transmit. Also, I can take advantage of sleep modes. I can simply increase the transmit power if I find my walls attenuate the signal too much. Finally, when DigiMesh comes out, I will still be able to try it out since I am using the Series 1 modules.

The $200 extra spent, I think, will be well worth it.

Also, your comment about whether or not DigiMesh will prevail…I think it will because it functions as a true Mesh network (atleast in theory). No more coordinators, no more routers, just end points that can route. This is perfect for sensor/home automation/security networks that don’t require a lot of bandwidth.

Nullz

Nullz:
stevech,

Thanks for the helpful response. I was hoping you would comment. Your estimate range for the 1mW radios of 50-100ft through 2 or so walls has lead to my decision to purchase the Xbee-Pro 802.15.4 modules. After reading the product manual, I have found out that the power level can be reduced to 10mW via the PL command to reduce power needed to transmit. Also, I can take advantage of sleep modes. I can simply increase the transmit power if I find my walls attenuate the signal too much. Finally, when DigiMesh comes out, I will still be able to try it out since I am using the Series 1 modules.

The $200 extra spent, I think, will be well worth it.

Also, your comment about whether or not DigiMesh will prevail…I think it will because it functions as a true Mesh network (atleast in theory). No more coordinators, no more routers, just end points that can route. This is perfect for sensor/home automation/security networks that don’t require a lot of bandwidth.

Nullz

Re DigiMesh - it’s great that Digi did this because of the big flaws in ZigBee. But - keep an eye on the politics: XBee series 1 use the FreeScale chipset which originally used the Figure8 ZigBee stack. T.I. bought Figure8 so this put that ZigBee stack at odds with T.I. who competes with Freescale. So it seems that series 1 was realigned to be '15.4 only and then came Digimesh to deal with those that did not want ZigBee due to the license costs and/or the lack of time synch’d rendevous (needed for battery powered/sleeping mesh routers). The reason for pause on DigiMesh, in the long run, is the politics with Freescale, that Digi is promoting the Ember based Series 2, and the competing NWK layers from ISA SP100a and 6loPAN.

But gee, in IEEE 802.15, there is a lot of dynamics and turn-over. The good news is that there are so many chip vendors and module vendors who have paid the $$$ for FCC (or equiv.) certifications internationally - mostly Digi/Maxstream. Zensys watches all this too, as they kind of captured the home automation meshing business with Z-Wave, whilst ZigBee languished in committee. So now ZigBee has bet the farm on capturing a good share in the automatic meter reading and instrumentation “smart grid” power industry.

I’m not well connected enough to know if DigiMesh is a contender in all of this; I would guess that the meter manufacturers want a proprietary network layer/messaging to discourage competition. The electrical system operators may half-heartedly want standards.

It’s all pretty political.

This is an old thread but I would like to join in. If you are looking for good indoor range you need to use sub 1Ghz frequency. I don’t see that Xbee has that type of devices, I know that Synapse Wireless has their range of Snap devices that work in 868/915 Mhz (EU/USA) so that would be my choice… and Synapse modules work setup automatic mesh network…

Are there any other good suggestions for indoor modules?

Digi/XBee makes sub-GHz.

My suggestion is to use the higher transmit power modules for indoor for a 1800ft. or more home. Yes, the transmitter power consumption, if battery powered, is an issue. But these network nodes should transmit very infrequently. The higher power reduces the number of retransmissions for error correction which in turn reduces battery consumption.

IMO, 2.4GHz should be used rather than sub-GHz, for indoor; because of lower cost (higher manufacturing volume for 2.4GHz = lower prices); few vendors for sub-GHz; North America and the EU have different frequency band standards, etc. The antenna size of sub-GHz is much larger than 2.4GHz due to the wavelength. If a too-small antenna is used in sub-GHz, the 6dB advantage that 900MHz has is lost to crummy antenna performance - often negative gain with teeny antennas.