In the day-to-day competitive battles between smart meter Neighborhood Area Networking (NAN) and radio technologies, discussion of the “best” transmission frequency is often at the forefront, and for good reason. Transmission frequency influences range, propagation through objects such as foliage, and the communications data throughput. So the win by Trilliant at Maine Central Power (CMP), announced last week along with a big $106 million investment by ABB, GE, and others, is especially interesting.

Trilliant has been a bit of a lone wolf in the use of 2.4 GHz for their RF mesh NAN, based on an IEEE 802.15.4 standard-based MAC/PHY protocol layer. Virtually everyone else, including Silver Spring Networks, Itron, Landis+Gyr, and Elster, use proprietary MAC/PHYs in the 902-928 MHz frequency range. Why? Because according to the physics of the Friis Transmission Equation, all else being equal, communications will travel farther at 900 Mhz than 2.4 GHz. The “2.4G vs. 900M” arguments go well beyond the NAN; a quick Google will get you similar battles over cordless phones, headsets, and other applications. This is also a fixture of the ZigBee vs. Z-Wave war for wireless home automation.

The issue is that the “all else being equal” caveat is rarely true in practice. Radio transmit power, encoding scheme efficiency, receiver sensitivity, noise rejection characteristics, and antenna design are all major factors in real-world performance. A slightly more robust form of Friis’ equation (see figure below) contains enough variables to make one think it really is rocket science. A major benefit of 2.4 GHz is the possible increased data bandwidth. For example, the IEEE 802.15.4 MAC/PHY used by Trilliant has a link data rate of 250 kbps, while the competing 900 MHz offerings range from 19.2 kbps to ~150 kbps – a significant difference.

What makes Trilliant’s win at Central Maine Power most interesting is the nature of CMP’s service territory, which covers about a third of Maine. Those outside New England may not realize that Maine is a very big place. CMP’s service territory is bigger than nine U.S. states, including each of the remaining New England states. With a relatively low customer density (only 600,000 customers), one would assume than NAN communications range was a major decision factor.

So the lesson may be that NAN evaluation should not stop at Friis. There are many other considerations, including potential interference from other unlicensed users at that frequency, immunity to weather, operation in a multipath environment, mesh protocol robustness, and radio design characteristics, that should be included in any assessment. One thing is sure – the arguments will continue to be made on both sides, and real-world experience will trump fancy equations every time.