News | November 11, 2008

Wireless Sensor Networks 2009-2019

By Peter Harrop, Chairman, IDTechEx

The dream is to automatically monitor and respond to forest fires, avalanches, hurricanes, faults in country wide utility equipment, traffic, hospitals and much more over wide areas and with billions of sensors. It will become possible thanks to Wireless Sensor Networks (WSN) otherwise known as Ubiquitous Sensor Networks (USN).

This article shares some of the research carried out for the new IDTechEx report "Wireless Sensor Networks 2009-2019" WSN can assist with many of the big issues of this century including security, safety, protecting the environment, animal and plant conservation, healthcare and efficient food production for example. Meanwhile, 99% of sensors installed in the world are still wired and, over the next ten years, WSN will constitute no more than ten percent of the wireless sensors that are sold, mainly because of technical challenges.

There is much to go for. WSN already beats the wired alternative with 90% reduction in cost, even at current prices, and often WSN enables things to be done that are otherwise impossible. Little wonder that hundreds of research centers are pursuing the subject and there are already well over 100 suppliers of this or that part of the jigsaw puzzle. An early objective for most schemes is wireless replacing all or almost all wiring, not just power and not just control.

WSNs are wireless mesh networks that link tiny devices that contain a wireless transceiver, a communication protocol handler, a microprocessor which is often application specific to contain cost and has appropriate firmware and read-only ROM as well as random access memory (RAM), an onboard power source which is usually a battery, and at least one sensor for measuring anything from temperature to humidity to motion, and/or an actuator. These nodes may also have a human interface such as a display with driver. Called Third Generation Active RFID, WSN involve tags (nodes) with more memory - typically 1MB to tens of MB and microprocessors and sensor etc that consume more power than traditional RFID. WSN sometimes subsume Second Generation RFID which is Real Time Locating Systems, though not yet with the best locating performance.

Starting with humble applications such as meter reading in buildings - the first killer application earning profits for several participants and strong growth - WSN will grow rapidly to become a $1.75 billion market in 2019. That will make them a significant part of what will then be an RFID business of over $25 billion, more than one quarter of that expenditure relating to active RFID where the tag has a power source. The term Wireless Sensor Systems (WSS) tends to be applied to more primitive wireless control of sensors as well as the mesh networking option. The WSS market is multi-billion dollar already.

Meter reading WSN are newly successful because the need is now recognised for users to have more visible and timely information on their power consumption and the mesh network is now an affordable solution. The nodes are relatively simple and are often made in China to achieve a selling price to the installer of only ten dollars or so, about one tenth of the average price of WSN nodes elsewhere where the complexity and ruggedisation are usually greater and volume sales are not yet forthcoming. Indeed meter reading nodes do not employ sensors in the normal sense of the word other than on old meters. The leading manufacturers of meters such as Landis & Gyr are keen on this opportunity to increase their sales. These leading utility meter suppliers now adopting WSN ship around 100 million meters yearly and, within the decade, we expect China to become a significant user as well.

Currently, Victoria in Australia, California, Moscow, Singapore and Sweden are most active in purchasing WSN nodes for meter reading but the UK and France are becoming active, with France planning legislative push. In France and some other countries, the ZigBee protocol - currently most popular for WSN overall - may not be favoured. The price of meter reading WSN nodes will drop rapidly as shown below, their eventual commoditisation being more than compensated by systems for other applications involving more complex, valuable nodes performing more sophisticated functions.

As Metcalf's Law puts it, the value of a network increases as the square of the number of nodes. WSN with up to one trillion nodes in a given system - applied as paint or adhesive dust or mixed into concrete, road surfacing etc - should be possible one day but currently only systems with up to thousands of nodes are commonplace, primarily because the batteries in the nodes cannot cope with frequent interrogation or long range as desired in most of the larger potential deployments and - related factors - the nodes cost too much and are not sufficiently reliable to be embedded for life or otherwise rendered inaccessible for decades. Ten years ago, the US Military conceived smart dust that could be dropped on and around the enemy in trillions that would monitor its actions and communications and be effectively indestructible. That remains a long way off. However, companies such as Arch Rock have announced very widely deployable WSNs in 2008.

Hot territories

The USA dominates the development and use of WSN partly because of the heavier funding available there. For example US government funding of WSN research at US universities was over $100 million between 1998 and 2005. The CENS project at UCLA started its WSN work in 2002 and it has $40 million over 5 years for this work. Other countries invest less but Korea is the only country with a national WSN program. Of 141 developers and suppliers profiled in one table of the new IDTechEx report "Wireless Sensor Networks 2009-2019", the geographical distribution is as follows.

In WSN mesh networks, any node can electronically talk to any other node provided it is in range. WSN nodes are self-forming - put them in the range of the network and they automatically join - and they are self healing - they find a way to signal around any failed nodes because the mesh of communicating nodes allows multipath hopping of the signals in and out of the network. To that extent they work like the internet but whereas the internet has TCP/IP as the standard transport protocol, WSN do not yet have a common stack on which all applications are implemented.

WSN are easy to install and exceptionally scalable - deployment of more and more nodes in a system is relatively easy - and they are responsive to their environment. They are designed to provide information of great integrity and usefulness from large numbers of imperfect, low cost nodes infrequently sending and receiving modest amounts of information to each other over modest distances, many short ranges adding up to long range. Sophisticated software and firmware and large numbers of affordable nodes are key - the opposite of traditional RFID and wireless sensing. The nodes can even be reconfigured remotely to perform a different task. Some refer to WSN as Third Generation Active RFID but RFID is already a business of more than $5 billion, so why have WSN sales been so slow to take off?

Power challenges

Some nodes have batteries go down in only two weeks and others lasting two months are often a problem. Larger batteries than those common in conventional RFID can be used to give a more conventional 3-10 years life but that is not enough to let them be embedded for life, as is desired in some applications. Work proceeds to make WSN less power hungry and to supply more power, for example by energy harvesting and improved laminar batteries.

Node numbers

In many envisaged applications, billions and even hundreds of billions of nodes need to be deployed but the favourite protocol ZigBee is limited to about 65,000 nodes at present and it is rarely economic to deploy more than thousands at current node prices. Therefore new protocols are being developed and printed sensors - and ultimately everything printed - should get node price down.

Other limitations being tackled

Other technical limitations are also being tackled. For example, today, WSN have highly constrained resources such as processing, storage, bandwidth, power and causing and suffering from interference. Use of Ultra Wide Band to a new standard IEEE 802.15.4a is promising as one way of improving the situation. Sensors, batteries and antennas are increasingly printed and there is a clear roadmap to printing WSN logic and memory as well. For progress in printed and potentially printed electronics attend Printed Electronics USA, San Jose, December 3-4