Methods designers can use to enable low power radio communications

8 mins read

Short range, low power radio is a hotbed of anarchy amid the highly regulated sections of the radio spectrum, reserved for access by the military, communications companies and TV channels.

The low power radio zones are tiny slices of spectrum where, as long as you obey the rules of transmission power, you can do more or less what you like. And, for decades, most companies have – choosing to define and operate their own protocols. However, in the past 10 years, standardisation efforts have accelerated with the aim of expanding the market for low power wireless and bringing disparate electronic systems onto the internet. If there is one winner in this low power radio world, it is IEEE802.15.4. But that is largely because it benefits from the effect that Andrew Tanenbaum noticed when he quipped about standards. "The nice thing about standards is that you have so many to choose from." There is no single radio protocol in IEEE802.15.4 – it is a collection of physical layer interface specifications and media access layer control protocols that can be mixed and matched freely. Since the release of the first version of IEEE802.15.4, which itself contained several options for the physical interface, the IEEE has followed with three updates that double the number of possible types of radio link. The original standard was based around three unlicensed bands. For Europe, it supports the 868MHz band, for the US, it's the 915MHz band. However, for greatest international support, most choose the increasingly congested 2.4GHz, which has to be shared with Bluetooth, WiFi and a variety of other unlicensed radio systems. The original version of IEEE802.15.4 described the use of direct sequence spread spectrum (DSSS) techniques for each of the bands – running at up to 40kbit/s in the sub 1GHz region up to 250kbit/s in the 2.4GHz region, thanks to its greater bandwidth. Three years later, data rates for the sub 1GHz bands were increased and the revision introduced other modulation schemes, allowing the use of binary or offset quadrature phase shift keying or a combination of binary keying and amplitude shift keying. More recent work has opened bands for use in China – centred on 315MHz, 432MHz and 784MHz – and in Japan, with a 953MHz band. That's not all. A decade ago, the 802.15.4 working group voted to establish a study group that would look at adopting a physical layer that used alternative types of radio transmission to traditional shift keying or spread spectrum techniques. One method proposed chirp spread spectrum for the 2.4GHz band instead of the conventional direct sequence or frequency hopping techniques. The other was ultrawideband (UWB). Because it uses the spacing between broadband chirps to relay data, rather than modulating a bit pattern on to a narrowband carrier, UWB is not tied to a specific frequency band. In practice, UWB communications are limited to regions above several gigahertz or to sub 1GHz frequencies in order to avoid interference with sensitive spread spectrum protocols – such as the Global Positioning System – which also spread radio power over a sizeable bandwidth. Approval has been granted for UWB at more than 6GHz in Europe, with allowance for operation below that range at very low power levels, and between 3GHz and 10GHz in the US. However, operation in these regions tends to limit range and makes transmission more dependent on line of sight, which causes problems for some of the anticipated applications. When the UWB version of 802.15.4 was being developed, the rationale for yet another radio interface was that a number of companies had identified problems with existing RF tag technologies. In a warehouse, the tags let you know there are pallets in range with certain contents, but a forklift operator armed with just a scanner will not be able to work out which one is which without checking each pallet individually. A network of three or four UWB-enabled receivers can use precisely measured time differences to pinpoint each transmitter and send that information to the operator's terminal. At low data rates – typically less than 200kbit/s – the potential range of UWB is 100m or more. However, operation at frequencies well into the gigahertz range will limit the potential range and push up the number of receiver nodes. Even so, if the UWB version of 802.15.4 takes off, it might lead to a TV remote control that tells you that it's fallen down the back of the sofa. One approach to the RF physical layer that is not covered by any part of 802.15.4 is the concept of white space radio. Frequency hopping protocols such as Bluetooth already use a limited form of white space radio – constantly shifting the transmitting frequency to 'hop' out of the way of interference. White space radio opens the concept to a much wider range of the radio spectrum, looking for gaps in transmission both in frequency and time, using those for communication until an interfering station starts broadcasting. White space radio is still in its infancy, although it has been bolstered by the formation of the Weightless group, which numbers ARM, Cable & Wireless and Neul among its members. The group is working on a protocol that can be used from short ranges to up to 10km to support applications such as remote meter reading. With 802.15.4, the question of which standard to use gets worse as you move into the transport and application layers of the protocol stack. A baffling array of companies and groups have opted to use 802.15.4 as the substrate for their own wireless networking protocols. The most famous right now is Zigbee, which has captured a large number of supporters, but has yet to translate that support into a market that rivals more widely used protocols such as Bluetooth and WiFi. According to IMS Research, Ember – acquired earlier in 2012 by Silicon Laboratories – was the market leader in Zigbee chips in 2010, with a share of around 30%. By the end of 2010, the company had shipped 10million chips since its inception a decade earlier. That figure had grown to 25m by the time of the company's acquisition. According to IMS, wearable devices alone – many, such as sports watches, using proprietary low power radio protocols – saw shipments rise in 2011 to 14m a year. Extend that to mobile phones with Bluetooth and the number rises to hundreds of millions. The Zigbee Alliance and its members promote the standard as being the driver for the 'internet of things', linking building automation controls and personal devices together wirelessly. But it has to compete with other specifications, such as 6LoWPAN; an adaptation of the core internet protocols for low power radio which also uses the 802.15.4 stack to provide the physical and media access layers. Because the existing internet is running out of addresses, 6LoWPAN takes advantage of the much larger IPv6 address space to make it possible, at least in principle, to give an IP based ID to every electronic device in the world for years to come. A contender with its roots in the industrial market – and therefore a large installed base of equipment that already uses its wired equivalent – is WirelessHART. Dust Networks, one of the developers of the concept of the internet of things and recently acquired by Linear Technology, developed the time sequenced mesh protocol (TSMP) behind WirelessHART. This protocol is used for the MAC layer in place of those defined in 802.15.4. Another protocol that has emerged in industrial process control is ISA100, which also employs TSMP. The attraction of protocols such as TSMP is power consumption. With most protocols, the problem with having a receive channel is that the node does not know when a message will turn up and simply has to keep listening. This makes it hard to power down, which is bad news for a wireless sensor node, where long periods spent asleep punctuated by bursts of activity, provide the key to a 10 year service life on a single battery charge or make energy scavenging viable. Some applications can work around this problem by simply not listening to the radio environment. In Spring 2012, the IEC approved a standard based on EnOcean's low power communications protocol which revolves around this concept. The protocol is used in devices such as wireless light switches – force on a switch with a piezoelectric backing provides just enough energy to send a command over a short range wireless link. Although the instantaneous power draw of a transmitter is usually higher than that of an active receiver, the long term average consumption is usually much lower because the radio link only activates when there is data to communicate. Having a system that can only transmit presents some configuration issues – it's hard to verify automatically that a switch is still working remotely without sending someone out to test it physically. One option is for a node to only listen when it is expecting to receive data, one of the philosophies behind protocols such as TSMP or the time synchronised protocol used by Zigbee. Nodes that employ the protocol can synchronise their clocks using received data packets to within 1ms to ensure there is not too much overlap between periods when two or more nodes are active. The time slotted nature of the protocol also reduces the probability of collision if more than one node attempts to transmit data, harking back to the time slotted version of the Aloha protocol used to connect computer users on the Hawaiian islands in the 1970s and which provided the inspiration for ethernet. One mechanism that promises greater energy savings, but which is not yet mainstream is the concept of event driven radio. This uses a second, very low data rate RF signal to alert nodes when a transmitter wants to contact them. Only then do those nodes activate the receiver for the main RF link to listen for the data. Once the data has been received, they put the main receiver to sleep, allowing just the event driven radio to function. Research institutes such as IMEC are continuing work on event driven radio and one of IMEC's designs has a power consumption of 50µW, ten times lower than that of the analogue section of a typical narrowband receiver (see fig 1). IMEC's receiver dispenses with many of the analogue blocks common to just about every other RF processing unit in use today. Oscillators and phase locked loops consume too much power and so have been excluded. The modulation is simple on-off keying because this allows the receiver to use simple energy detection to decode data. Using some passive components to tune the detector for the target frequency band, the received signal is amplified and passed almost directly to an A/D converter for downsampling and processing in the digital domain which, thanks to the low datarate, can be handled by low energy logic. Because the receiver wakes up almost as soon as something happens, rather than wait for a time slot to roll around, the average response time is much better, providing greater energy savings over conventional radio systems if an application is latency sensitive. Event driven radio (see fig 2) may solve one of the problems that faces a number of the low power radio protocols. A further question is that of network structure. Protocols such as Zigbee allow for a mesh structure – theoretically more robust and easier to manage than classical network designs such as stars and rings. The mesh structure puts more intelligence into the nodes themselves so they can find an effective route for a packet to reach its destination through an arbitrary mesh. Zigbee uses a technique developed at the Nokia Research Centre, University of California, and the University of Cincinatti. Known as ad hoc on demand distance vector (AODV) routing, the approach only looks for a route between two nodes when one wants to send data to the other. A similar protocol has been proposed for 6LowPAN. These approaches are unlike IP networks, in which routing tables are predefined and updated at regular intervals using specialised route discovery packets. With AODV, a node broadcasts a message asking for routes when it needs a connection (see fig 3). This results in a flood of responses. Nodes with a known route to the destination relay their proposed directions to the node that made the requested through the mesh until they reach the requester. The requesting node then picks the one with the lowest number of hops and caches it in order to avoid having to go through the same procedure for each packet it sends. The AODV protocol is relatively simple, but can be problematic in terms of energy consumption because it demands responses from many nodes when setting up a route – and nodes have to listen constantly for those packets. The Zigbee group is working on a low power version of the network to allow use with energy harvesting nodes. These nodes will not be full members of the network, but will use other nodes as proxies for routing data into the network to avoid them having to participate in these activities. Some groups are using wireless networking to develop more exotic forms of routing. For example, MyriaNed uses the idea of human gossip to disseminate data across a network. Nodes talk to their neighbours in the hope that the data will eventually find its way to the intended destination. That networks such as MyriaNed are being developed show how far low power radio has to go in terms of development and standardisation. While proprietary protocols are beginning to give way to more widely supported standards, it will take a long time before there is a clear winner – if that ever becomes the case.