GENERAL PROJECT FINDINGS

1. Bi-directional data radios show great potential for supporting current and future Environmental and Biological Field Science.

2. In particular spread spectrum digital radios manufactured since 1990 - since FCC rules permitted their manufacture, sale, and use - operating in the FCC Part 15 Bands, hold such potential that the way is opened to revolutionary ways field data can be collected, transmitted, acquired, and distributed world wide, not only to institutional research scientists, but also to educational populations from grade school through graduate school, and to any citizen in the world who has a web browser and access to the Internet. Even low bandwidth access. (dial up)

3. Several characteristics of both radios and attached devices - from individual sensors and data loggers, computers, and networks - combine to stimulate and reinforce this revolution.

a. Data loggers are now routinely and very widely used to store locally connected data.
b. Increasingly the stored data is in standard forms, which can be made interoperable.
c. The sensor industry has become quite ubiquitous. They are capable of being integrated into data logger, storage, or processing systems,
d. Low cost, processor driven, sensors handle data in a wide range of forms - numeric, images, sounds, increasingly stored as digital data.
e. Data radios can deliver data, error free, over increasing distances, through hostile environments.

4. Current commercial satellite systems and services can readily be interfaced, affordably, to digital wireless devices radiating out from the base station, making data collection wirelessly possible from virtually every place on earth, no matter how remote from urban communications infrastructures.

5. Defacto standardization which has spread since the Internet became popular, - - RS232 serial, Ethernet, TCP-IP, 802.11b - makes it relatively inexpensive and easy for technicians supporting science data networks to assemble, interface, configure and deploy wireless data networks from off-the shelf equipment, and connect up their data loggers and sensors to data bases and to all researchers via the Internet.

6. Radios operating in the 902-928Mhz unlicensed range are superior to radios using the 2.4ghz, or 5.8ghz frequencies for fetching field data through vegetation and trees.

7. Current FCC power rules for the US (radios and their associated amplifiers and antennas cannot exceed 36dBi), which do not differentiate between Urban and Rural or Remote-area uses, do not adequately support field science. While off the shelf data radios operating under current FCC rules, can, using costly relay stations, be made to operate from 1 to 10 miles even through wooded areas, there is no reason from the standpoint of 'interference' with other radios in remote areas where very few, if any such radios are in use, why the FCC could not, or should not, grant permanent increases in power (to at least 10watts or 40dbi) for rural uses. While it is possible to seek waivers under Part 5 FCC rules for 'Experimental Licenses' and there are costly families of License-required radios, these should not be necessary.

8. Environmental and Biological Science Institutions/Universities are almost oblivious to the role and importance of the FCC in enabling, or inhibiting their uses of wireless for their field science. There simply is very little 'wireless' expertise in these institutions. Thus there is little knowledge of spectrum rules, how to find them on the FCC web site, how to comment on FCC proposed rule changes, or how, collectively influence FCC rulemaking, even though the process is open and public and all involvement by universities or researchers or their professional organizations can entirely be by email and web access to FCC government nets.

9. Likewise American field science is not even on the radar screen of the FCC as an important public constituency which should be served.

10. There is a need for universities which have substantial environmental or biological science programs to add a wireless component to their IT staffs. While individual Science Projects - such as the NSF funded LTERs - or the work of the 150 Biological Science field stations can recruit or subcontract with wireless experts to support a given project, this is no long term solution for an institution. For deployed wireless networks supporting long term ecological research projects will have to be maintained, troubleshot, and/or modified long after temporary wireless experts are gone. In four of the five LTERs worked with by this NSF project, if the technicians who mastered this new field of digital field wireless leave, there is little likelihood anyone will replace them. Just as the use of personal computers, then modems, then IP networks had to be learned over the last 20 years, both at the individual user, and the institutional installer level, so will wireless be gradually learned by all. But the art of making valid technical site surveys, deploying wireless networks, troubleshooting them, and interconnecting them over Internet works is not trivial. At this time it is still specialized work. This is not to say that technicians in the LTERs supported by this Wireless Project have not learned a lot by observation and working besides this PI and colleagues, but none of them are yet at the larger scale, institutional-level of wireless technical expertise, management, or policy making.

11. Reliable delivery of power to radios in the field - through all seasons - is the largest obstacle in the use of wireless devices. A large proportion of data loggers are emplaced in valleys, woods, and low land places - exactly the places which solar panels deliver the least efficient power to storage batteries. In cold regions or where winters are cold, it can cost more for solar or wind power generation and battery storage, than complete radio systems. Much development needs to be done to overcome this problem. Harvesting power from the field, fuel cell technology are among the most promising technologies which need to be explored.

12. Since wireless has proven its ability to serve Data Loggers, but also since data logger systems can be costly - at least $3,000 per site, and off the shelf radios with their associated power and battery storage systems can add another $2,000, such $5,000 complete systems do not 'scale' well. Repeatedly PI's wished that wireless could be used to deploy many more systems to monitor the environment. The answer lies in the development of totally new small radios, singular sensors, rechargeable batteries, which can operate in a mesh network (fields of sensors), and the decreased use of more costly data logger plus radio installations (except for key location meteorological stations with many sensors)

13. It is more costly in energy to 'communicate' than to 'process' data collected. Paralleling the development and use of smaller, short range, networking radios, is the need for development of very small processors to connect individual sensors and the radios. So that data collected can be processed and perhaps reduced to the least amount of significant data, and only that small amount communicated wirelessly. Conserving power over long periods. The typical combination used in this project of off the shelf radios, standard large capacity data loggers, and complete power systems - which can last at least 6 months untended in severe climates - is not necessary, and from a dollar cost standpoint, undesirable. Larger numbers of smaller data collectors better serves field science while remaining the most cost effective.

14. There is an accelerating rate of change and development in wireless data radios. Between the beginning of this project in 1999 and its end three years later, there has been explosive development of new radios, new amplifiers, new antennas, 'smart' radios, and more complex signal processing methods. Costs have plummeted for some radios (802.11b) are reasonable for some new ones (5.8ghz spread spectrum, UNII band radios), but the industry is stagnant - in pricing and development of radios operating in the best frequency ranges - 902-928mhz - for field science. Again, this is because manufacturers do not see field science as a market (even though one Data Logger company, Campbell Scientific has a growing market and is profitable for its support of field science.) The Environmental Science community is not defining what it wants in wireless and communicating to manufacturers. Unfortunately it is taking whatever industry gives it.

15. The vast majority of collected field environmental data collected from sensors requires low data rates to communicate efficiently. It is rare than wireless data speeds need to be over 9,600 baud. Even though one of the best radios for data collection - the Freewave serial port radio - can operate at 115kbps, none of those deployed by this project had to operate above 9,600 baud. Thus the high bandwidth radios - such as 802.11b 11mbps (half duplex) devices - have limited utility for field data collection. And such radios, which operate at 2.4ghz or 5.8ghz cannot penetrate vegetation nearly as well as slower data rate, frequency hopping 902-928Mhz radios. The only exception to this finding, is in the case of deploying video cameras for continuous broadcast over wireless nets in some field science locations. Then higher data rate radios are useful.

16. The most important task which has to be performed before a wireless network supporting field research is designed and deployed, is the technical Site Survey, in which either Spectrum Analyzers are used to determine link requirements and robustness of signal, or actual radios are used, with a variety of antennas. This aspect is the least appreciated or practiced except by professional installers. With experience technical staffs can make installations without owning or using costly spectrum analyzers. But a sound site survey is vital in all cases. Especially this is true if a network is to be set up, which might be later expanded. The basic layout has to be right, to permit such expansion without unnecessary reconfiguration of existing radios, antennas, or their mountings.

 

TECHNICAL PROJECT FINDINGS

1. The serial port, Freewave DRG series of radios operating at 902-928Mhz and using unlicensed spread spectrum Frequency Hopping protocols remain the best radios for establishing wireless links to and from data loggers in the field. In particular to and from the Campbell Scientific CRX series - which are the most favored data loggers by environmental scientists. These radios have the best receiver sensitivity, configurability, robustness, range, reliability, and tree and vegetation penetrating capability of any 915mhz radios tested.

2. 2.4 or 5.8ghz radios, from 100mw 11mbps 802.11b, either end user or access point radios from a wide variety of vendors, and at very low cost (under $100 for individual radio, or $150 for multi-user access points) are the least useful in gathering field data from data loggers.

3. 802.11b 11mb radios do have a useful role connecting up video or sound capture systems short ranges with clear line of sight, to long-haul radios or networks. Their interoperability and point to multi-point abilities permit them to be used in heterogeneous networks. They can be used the 'last half mile' by researchers carrying laptop computers to connect to the Internet, provided there is a long haul connection, wireless or wired, from the nearest access points.

4. 5.8ghz spread spectrum or UNII band radios, operating from 11mbps to 100mbps are primarily useful for back haul from field sites, tower or high point to highpoint, in the support of field research.

5. The severe power limitations put on unlicensed data radios by the FCC, even in the most remote field locations where 'interference' as experienced in urban or building complex setting is virtually nonexistent, severely cripples the ability of Environmental and Biological Research Scientist to do their field work supported by wireless networks they can set up.

6. The latency - from 500ms upwards of bi-directional satellite Internet services, and the power requirements of field ground stations - limits their utility to be used as a remote-area 'base station' for radiating out to the range limits of other wireless devices. This is particularly true for efforts to communicate over the Internet directly to the Campbell Data loggers - since the program time-out code in the Data loggers do not take into account long latency of connection to PC208W software running on remote PCs. This can be handled in the case of custom programs other than PC208W software designed to fetch data from the data loggers and store it in data bases.

7. Campbell Scientific has brought out their own unlicensed radios configured to best interface with their data loggers - such as interfacing with the modified Serial protocols called CSIO, and power - from the data logger. However these radios are modeled after 2.4ghz radios, have very low power (100mw or less) and because of the combination of their frequency and power must be used in pure 'line of sight' configurations. Which is rare in field deployment of most data loggers.

8. In spite of the explosion of sale and use of 'Wi-fi' (802.11b) radios which operate in the 2.4-2.483ghz bands in the general economy (reported as more than 1.5 million now sold a month) with corresponding lowering of price through competition to below $100 a radio there has not been a corresponding increase in the manufacture or sales of radios which operate in the 902-928Mhz range, even though such radios are far and away the best suited for field science data networks. Accordingly 903-915Mhz radios still cost in the $1,000 or higher range. Thus there is a problem of 'scale' in the use of these costlier radios for field science.

9. In most cases - the off-the-shelf radios, either 915Mhz or 2.4Ghz types, are 'overkill' for most field science needs. They consume too much power, are capable of data rates higher than needed, have too many features, and cost too much. For such real world needs as linking 44 light sensors, costing $20 each, 150 meters to a data logger or back haul point, as one Puerto Rican research project needs, deploying all that is required to power and maintain $1,000 radios is not sensible. There is a great need for smaller, cheaper - in the $100 range complete with processor and power supply - shorter range radios that can link individual or a small group of sensors to the net, using 'mesh' networking techniques. It is hoped that later NSF funded Wireless Projects will research and prototype such useful digital radios.

10. Because of power requirements to keep field radios 'always on,' a technical strategy of either having a device, such as the programmable data logger to which it is attached, turn on and turn off the radio periodically to conserve power - or the addition of substantial, with ample margin, solar panels and 12 volt Marine quality gel cell batteries are needed.

11. The most useful radios for field science work can operate on 12volts, DC, because that is the most readily available in portable, battery with long amp hour storage form. And such radios can be powered from cigarette lighters in vehicles - which are needed in particular during 'site surveys' prior to selecting the places and configurations of wireless devices.