Weather systems

Weather conditions constantly change, making weather monitoring important. Today's weather monitoring systems are a bundle of high-tech instruments, including smart sensors with the capability to speak directly to digital systems (data loggers and computer stations). However, their usefulness would be limited if the systems could not be placed anywhere outside and in large numbers to collect more data. Here, the new wired and wireless technologies come into play, and the neighborhood weather stations can become a wired, wireless (802.11) or mobile network.


The generalized set of parameters collected or measured by a weather system are ambient temperature, wind direction, wind speed (anemometers), humidity (hygrometer), barometric pressure, dew point, precipitation (rain gauges) and global radiation (solar energy). There is nothing unique in the way the computerized weather station is built. The most advanced piece of the equipment is the smart sensor, which outputs not an analog voltage as in the good old days, but digital values corresponding to the value of the parameter being measured. So, for the temperature, a digital thermometer is used. The wind direction is detected by a vane with an attached magnetic rotor, which passes over magnetically activated switches to give the eight or 16 compass points.

For the wind speed, the solution is to use wind cups with a rotor featuring mounted magnets incrementing a counter (again, a digital solution). The more advanced solution without moving parts is a Doppler anemometer, which measures the speed and frequency of the sound. The smart sensors speak to a weather station integrated microcontroller through a one-, two- or three-wire interface. (Other implementations are possible.)

The timing of the procedure for collecting the data is given by a real-time clock chip, which is also responsible for attaching the respective timestamps to the measurement results. As a conventional microcomputer or microcontroller system, the weather station board also includes a memory — nonvolatile electrically erasable programmable read-only memory (EEPROM) or flash — to store the data for when a communication channel to the network becomes congested or unavailable. The heart of this story involves how the information is transmitted to the place where it's analyzed and a conclusion for the weather is made.

The wired way

Most of today's commercially available weather stations have built-in RS-232 or Ethernet LAN interfaces. So, it is obvious that they can be easily connected in a point-to-point topology (RS-232) or to a LAN infrastructure. The good news is that in this case, the power supply to the weather station can be arranged as Power over Ethernet (the IEEE 802.3af standard). The drawbacks are the necessity to run cables to the equipment and, as a result, the constraints in placement and distances covered.

The wireless way

The most convenient solution today is to equip every weather station with a V.90/GSM modem with RS-232 or similar interface to the sensors, and polling the stations for weather data via SMS. (See the mobile phone network arrangement in Figure 1 on page 38.) Another option is to use a general packet radio service (GPRS) network to send the data from the weather station loggers.

Some weather stations transmit their data through a dedicated spread spectrum radio modem and directional yagi-beam antenna, reaching a range of more than 30mi in the line-of-sight and above the tree line arrangement. The most sophisticated weather stations employ satellite telemetry technology, so they can be placed anywhere on the globe.

The smart weather stations (or even autonomous sensors) can be organized as an infrastructure 802.11 network. Such systems typically have short ranges of up to 400m for the 802.11b/g technology.

The limited range of the 802.11 solution can be resolved by the inclusion of a TCP/IP client in every weather station controller. This allows the unit to obtain an IP address (through Dynamic Host Configuration Protocol, or DHCP) and act as a network node visible from the whole world. In this case, the access point of the local wireless network should be layer 3 (network layer) connected through a router to another area network with Internet connectivity. (See the WLAN globally connected arrangement in Figure 1.) The sensor data can be monitored on a standard Web browser interface or can be linked to database applications. The calibration and control of the station is performed using a Java-based interface in this case.

As the information gathered by the unmanned weather stations is extremely important, often a redundant link reservation is accomplished. For example, one of the stations in Figure 1 has both GSM modem and WLAN interface. This way if one of them (or the respective network) fails, the connectivity to the station data is not broken.

The mesh networking implementation

The more advanced mesh network can also be used for wireless weather stations, and the topology exhibits some analogies to the well-known ad-hoc networking. (See Figure 2.) The advantage of such a solution is the ability to expand the sensor network with new weather stations without the need to reconfigure the network infrastructure. The only requirement is that a weather station must be in the RF range of another station so it remains visible to the network.

The mesh network exhibits two features that are important for the reliable transfer of weather data: It is self-configuring and self-healing. It is the perfect solution for areas without network connectivity at all — mountains or uninhabited regions — the domicile of the typical weather station. The traffic model for a mesh weather station network differs from the conventional wireless network model. As the weather stations get closer to the data collecting node (base station), the more the many-to-one traffic approximation applies. Many packets from different nodes arrive to the base station closer node to be relayed. This means that the stations closer to the collecting node are congested with traffic and their controllers and wireless cards shall be able to process the packet flow. It is even possible for the collecting node to be a moving object — an airplane flying over the area and collecting the data from the relay nodes.

Managing the weather network

The packets traveling the different types of networks are not only weather data, but also management information. Many of the sensors need to be set up, calibrated or firmware updated with commands for measurement tasks and schedule (polling of the stations) that need to be exchanged. All this is part of the management functions, which can be centralized or distributed among manager agents (located in the more powerful weather station nodes). The management information burden can be lowered by the use of the conception of the event-driven network. In simpler terms, traffic is exchanged only after a measurement by the weather station is made. From the security point of view, the management messages data need to be authenticated and encrypted.

Power supply limitations and solutions

All wireless solutions suffer the same main limitation — power application to the stations. There is no problem if a mains power supply is available at the location of the station. If no mains is present, one solution is to design and use power efficient protocols with battery supply. Another option is to power the weather station by solar energy panel, which makes its life unlimited (compared with a battery-powered limited lifetime instrument). The combination of reduced power sleep modes and efficient medium access protocols (MAC) improves the energy efficiency of the weather stations at the small price of somewhat increased delay in the response of the system.

Emil Vladkov is an associate professor at the University of Sofia, focusing on communication technology and digital signal processing.