You only need to visit a few web sites to discover that the current drive in technology is to make everything wireless. In the same way that the fractional horsepower motors revolutionised consumer product design in the sixties, followed by transistors in the seventies and microprocessors in the eighties and nineties, wireless is set to do the same in the coming decade.
Wireless has already transformed the things we carry and the way we work in the wider area, most notably with our eager acceptance of the mobile phone, both as an extension of our personal communication and also as an object of desire. The question is which wireless technologies will appear in the plethora of electronics devices we currently buy.
There is no easy answer – despite the claims that their marketing machines may make, no one wireless technology is the universal panacea – each has its specific niche within the brave new wireless world. Most have specific areas of excellence and many overlap. To choose which suits your need, you need to consider carefully what you want to achieve and the environment in which you wish to deploy it.
Machine to Machine or M2M connectivity is the buzzword. The market analysts Deloitte have estimated that today there are four machines in existence for each human being. By 2020, they extrapolate that to over thirty machines for each man, woman and child. If they could be made to talk to each other, the consequences could be immense.
It’s easy to look at this as science fiction, but already most of us in the developed world unwittingly use M2M wireless every day. We use it to control our televisions via remote controls, and we use wireless key fobs to unlock our cars. These are proprietary wireless solutions, made possible by the sheer volume of these everyday applications; what the new wireless standards offer is a raft of “off-the-shelf” wireless components, where the overall volumes will drive down the cost to the point where any manufacturer can contemplate making their product wireless. At which point wireless connectivity will become endemic.
Every application has its own, unique requirements, but the general shopping list of these remains constant across most wireless applications: Range, Data Rate, Battery Life, Security, Cost, Quality of Service, Interoperability, Qualification and Approval.
The relative importance of each will vary, but it’s unusual to find other parameters involved in the choice of radio. Ease of use would be a nice feature to add, but all too often that’s only considered after the event.
The first two, range and data rate generally provide the first selection points regarding which are the likely contenders. Although most radio standards are capable of transmitting small amounts of data, those that are designed for higher data rates are usually inefficient if you only need to transfer a small number of bits of information. However, if you already have an infrastructure for a higher data rate standard, it may be more cost effective to continue to use that for your local M2M connections. A classic example is the use of Wi-Fi or 802.11b for monitoring small numbers of machines when access points are already installed. Equally the choice of Bluetooth becomes attractive if the application needs to talk to a mobile phone or headset, as Bluetooth enabled telephony devices are so readily available.
Taking a broad view of range and throughput gives a rough pecking order of the different standards:
Whilst such a chart can direct a user or integrator to their first choice of wireless, there is inevitably a trade-off of other features. Of these, the most important are usually topography and power consumption.
Topography refers to the way in which radios can talk to each other. In most cases this is not a limitation of the radio itself. Radios are naturally promiscuous devices, that are happy to talk to any other radio of the same standard. It is the protocol stack that sits above the radio itself which determines the flexibility of connection scheme that can be implemented. Unlike a cable connection where the link is determined by the physical sockets on the two pieces of equipment that are joined together, wireless opens up a cornucopia of connection possibilities, from the simple to the complex.
The simplest scenario is straightforward cable replacement, where a wireless link replaces a fixed cable. Every wireless standard is capable of this, but there is a catch. Whereas a cable does not discriminate between voice and data, a wireless connection does. In fact, most wireless standards are only aimed at data transfer. The reason for the difference is that a transmission down a cable is rarely disturbed by interference, whereas a radio signal can often be delayed or corrupted. Such a delay is very obvious during speech or music, and to be a suitable carrier for these applications a radio standard needs to implement a feature called Quality of Service (QoS), which guarantees when data will be sent. Of the available wireless standards this is currently only supported within Bluetooth, which provides QoS through synchronous channels and guaranteed time slots for asynchronous channels.
Most commonly found in wireless LAN applications, infrastructure modes refer to a situation where an access point radio is connected to a backbone network, which may be wired or wireless. Other radio devices use this access point to gain access to the network behind it. Conceptually it is the wireless equivalent of the Ethernet hub. One fundamental aspect of an infrastructure mode is that devices cannot talk directly to each other, but need to communicate through the network behind the access point. Wi-Fi, Zigbee and Bluetooth all support infrastructure mode.
Multiple Connections or Ad Hoc
Multiple ad-hoc connections bring the benefits of promiscuity to a wireless network. Unlike a cable, a wireless device can start by talking to one other device in a cable replacement mode, but then bring additional devices into the conversation. It differs from an infrastructure mode in that devices can have a constantly changing group of connections and connected devices. Multiple connections can coexist alongside infrastructure mode and enhance the functionality of wirelessly connected devices, but more commonly are found in groups of mobile products that communicate with each other. Bluetooth is by far the most popular implementation for this ad hoc connectivity scenario, where for example a phone may connect to a variety of devices, such as a headset, a PC, a printer and a VoIP access point.
Mesh networks take the connectivity one step further and allow a core of devices to talk to each other. These keep track of which devices are connected and provide inherent intelligence of how data can move between them. When a device on the periphery of the network wishes to communicate with another device, the units within the network arrange the way the data moves from node to node. Mesh networks have some fundamentally attractive
advantages, most important of which is redundancy. If a particular unit within the core of the mesh fails, devices around it can reroute the data to its destination. The other, counterintuitive benefit is that the more radios that are added within a given area, the higher the potential data throughput. The cost of these advantages is a greater complexity and power requirement, which has currently limited mesh networks to specialised applications.
How long a battery powered wireless device will operate is one of the eternal debates within the industry. In the beginning, when the only two common wireless standards were Bluetooth and Wi-Fi the answer was simple –
Bluetooth was designed for battery powered, handheld devices, whereas Wi-Fi was a much higher power solution, predominantly for laptops which were regularly connected to the mains.
Over the past few years chip manufacturers have worked hard to reduce the power consumption of both Wi-Fi and Bluetooth still further and new standards such as 802.15.4 have appeared for extremely low power applications.
However, the actual battery life is not just dependent on the standard, but is closely related to the amount of information that is being transferred.
Wirelessly streaming videos will always consume immensely more power than an infrequently used application such as a light switch. What that means is that the average daily data transfer must be taken into account when looking at battery life. On top of that other features, such as mesh networking, may elevate a standard from low power per byte to high power per byte. So start by understanding the average and peak data flow for your application. More established technologies may fare better then newer ones. As standards mature, each new generation of silicon design usually results in an overall power reduction, so mature specifications may challenge more recent ones for power efficiency. For example, recent enhancements to Bluetooth, including the Version 2.0 + EDR specification are allowing it to target an even wider range of low power, battery based applications.
Despite the plethora of local radio standards, only two of them can today claim maturity, which are 802.11b and Bluetooth. Both are shipping in their tens of millions: 802.11b for wireless networking and Bluetooth as a cable
replacement within handsets and low power, handheld equipment. Between them they fulfil many of the requirements of short range wireless, and each is evolving to broaden their appeal and application. Others will arrive, but to be truly successful each new standard needs to find an application area where it will dominate. For most other standards, that is still several years away.
Finally never forget the option of combining more than one wireless standard. The different standards are often surprisingly good at co-existing, and the most effective deployment in terms of cost and power may ultimately be a combination of more than one.