The spark is lit

Wi-Fi both complements and supplants GSM in M2M communication according to Simon Taylor. Light, step back and await the M2M explosion.

For many years now, GSM has been the leading technology for remote machine to machine (M2M) communication. It has the merit of being relatively simple and reliable to install – but each link needs a SIM which means subscribing to the network – imposing a cost overhead that has limited the use of wireless connections.

For a great many applications, the wireless connection doesn’t need to be always on – it just needs to be available from time to time for convenience. For example, cold chain dataloggers that store temperature information for chilled foods in transit may need to comply with regulations that temperature data needs to be recorded, but the records don’t need to be accessed in real-time. Data could be downloaded over a subscription free short-range wireless link at the depot, or aggregated via a gateway to optimise the use of a GSM link back to base.

Up until very recently, WiFi has not been an option for embedded systems because of the overhead of TCP/IP and WLAN drivers. New modules have all of the protocols required to provide an ASCII based serial to WLAN interface suitable for connection to low-level processors. Most of these devices operate with well-known ‘AT’ commands, so are similar in operation to a dial-up modem.

Alternatives include other types of low-power radio, such as proprietary devices, DECT or ZigBee devices, but the lack of interoperable standards – even with ZigBee – tend to hold back their adoption. ZigBee suffers in particular from a lack of data exchange profiles, so although the radio layer is standardised, data interchange is not (yet), and a module from one manufacturer is unlikely to be able to communicate with one from another supplier.

Advantages of Wi-Fi

Wi-Fi has the distinct advantage of cheap, simple to install base station infrastructure, and an established connection to the Internet. Web applications and mobile hosted applications allow user-friendly interaction and monitoring on a handheld device, or remote PC. Web applications especially have the advantage that they are simple to deploy – just one push to a server, and the application is distributed to many – the ultimate in scalability. Soon, we will see remote control and monitoring on devices such as Apple’s iPhone either as web applications, or as an installed application.

The key is TCP/IP. The old model of peer-to-peer control and monitoring has the disadvantage of being complex and difficult to scale. TCP/IP however can employ multiple sockets – can be treated as virtual serial cables – and is generally limited by bandwidth. Scaling the system can be as simple as adding more ‘channels’ in server software, or at the extreme end, adding another server machine.

Back at the remote end (or ‘client’), we might have a simple temperature monitoring device for example. Let’s look at the options for getting data back from the remote station.

1. GSM. For a genuinely remote that has no surrounding infrastructure, that needs to provide data in real-time, this is still the best solution – it can be battery operated, and a little thought about data flow management can minimise the cost of attaching the unit to the network. Data could be sent back as an SMS, circuit switched data, or GPRS (Internet) data. The frequency and type of monitoring can dramatically affect the cost. Exception monitoring can be very cheap, but transmitting data every 30 seconds, say, will rack up costs very quickly! A cost benefit analysis will reveal the optimum frequency for communication. However, there is an unavoidable monthly cost to remain on the network.
2. If a ‘remote’ sensor is permanently on a larger site – or returns to base at regular intervals, using WiFi eliminates the need to subscribe to a GSM network. By either ‘piggybacking’ onto an existing WiFi installation, or installing a broadband connection, you can connect multiple clients into one broadband connection. So, in our example, if there were multiple temperature monitors in a 30 to 100 metre radius all of them could link into one WiFi connection. Using gateways and routers a WiFi infrastructure can spread over a very large site, with wireless devices all connecting into one main Internet connection.
3. GSM / Wi-Fi combinations. Taking aggregation a step further, we could use a GSM/GPRS device as the main Internet connection, but connect a WiFi device to our GSM modem to provide a gateway. Again, multiple wireless devices can connect to this gateway, making maximum use of one fee-paying connection, rather than using expensive GSM at every client.

Implementing embedded Wi-Fi

Looking at the sensor end, we could start with the Ezurio WiSM module provides a perfect easy to support Wi-Fi interface. This device includes a scripting language, allowing allows the module to read from its analogue inputs – in our example we are reading a temperature – and pass the data on under program control. In this way the module can log or send data back as required: at specified intervals or at a threshold for example. Within the API (Application Programming Interface), support is provided for the module to open TCP/IP sockets, which really are like virtual serial cables to the destination, and send simple ASCII data down the link. Other examples include packing the data up and sending it as an email, so the destination can be any computing or mobile computing device.

Thinking of the aggregator model, we just use the same client devices, but now instead of connecting to a normal Wi-Fi router, connect to a WiSM in host mode, itself connected to a GSM device like a TC65. This GSM/GPRS module can itself open TCP/IP sockets and thus work as a ‘bridge’ allowing communication from the remote WiSM device through to the Internet.

The WiSM has a web server on board, so can support on demand remote access.

Radio integrity

One of the challenges though, remains the integrity of the radio signal. All too often, this is neglected, with designers considering aesthetics rather than functionality. It’s good to hide an antenna away, but this will inevitably affect performance. The golden rules for antennas and good operation are height, clear air and ground-plane. While the designer may not be able to control all of these parameters, as much effort as possible should be made to give the antenna the best chance to transmit and receive a signal. The author has seen instances where the designer has complained of a device not functioning, where the antenna has been contained with the module within a die-cast enclosure!

Even in a correctly considered design, moving the antenna to the surface of a box, ensuring it is in clear air as far as possible, or even moving it away from the equipment, and up on a wall if possible will add many decibels of performance to the antenna. At the very least, consider a properly engineered solution such as Antenova if the antenna is to remain inside an enclosure.

Wireless and batteryless

Taking aggregation a stage further, other devices could be implemented for remote sensors, such as EnOcean’s wireless and batteryless transmitters. These ‘scavenge’ energy from the environment, for example, switches utilise the energy generated when pushing the switch to transmit a data packet. The STM 110 can use a solar panel or even a Seebeck effect device, which converts temperature differences to useful power, and can transmit up to three analogue measurements and the status of four digital inputs. Use these devices with up to a 300 metre range, into a receiver connected back to back to a WiSM, and again you have a useful gateway from remote sensors to the Internet. As a ‘demonstration installation’ TDC has implemented an installation using an STM 110 to monitor temperature, pump flow and water level in an ornamental fish pond. Not the sort of application that would be viable for most people with a GSM connection costing tens of pounds a month, but to have a Wi-Fi connected device into the home broadband that can send an alert email if some unforeseen event occurs would give great peace of mind to any fishkeeper with any concern for the livestock in his pond! Of course, this example transfers well to fish farmers, and then to any application where analogue or digital events need to be monitored. Remember, multiple sensors can connect to one Wi-Fi device, so cost is kept low, but functionality is high. Other devices that make connecting to Wi-Fi simple are the Mini iWi-Fi and socket iWi-Fi from ConnectOne, Multitech’s WiFi Socket, and devices from Wi2Wi, although all of these devices don’t have the processing capability of the WiSM, they are very capable serial to Wi-Fi devices, ideal when there is a host processor available. The ConnectOne and Multitech devices include TCP/IP, so the host processor does not to be Internet aware itself. The growing use of subscription free local wireless connections instead of or alongside GSM connections enable remote monitoring of a host of applications not currently regarded as economic. Personally, I am already monitoring a range of devices at home, and can even watch live video sent from my home on my iPhone. New, simpler to implement embedded WLAN modules are the spark that will truly ignite the long-awaited M2M explosion.

(Source www.cieonline.co.uk, August 2008)