Custom Search

Thursday, January 3, 2008

Mesh - The Emperor's New Clothes for WSN?!

I met Nick Hunn last Nov in a Bluetooth conference. I should not say "met" but "saw" or "observed". In that location, he was one of the prominent agitator of Wibree, now Bluetooth ULP. His enthusiasm, or over-enthusiasm on Bluetooth ULP can be felt 5% from the article attached below. You got to listen to his speech. He is a great speaker and marketer. I wonder he is carrying a wrong hat. Instead of CTO, he should be CMO. Interestingly, other guys from Nordic Semiconductor and Texas Instrument (Chipcon) are more conservative and were shaking heads during Nick's speech; and these conservative guys are the actual solution (protocol and silicon) providers.

I digress... I should write another blog for my thoughts about that conference, though two months passed.

Nick's article below does have one negative tone and it is towards mesh. So you know Bluetooth ULP does not support mesh. But his view makes sense and I can not agree more for many pragmatic applications today. Here is what he says:

"One of the most pernicious myths relates to mesh networks. Meshes have many admirable qualities when they are appropriately used. However, in many cases mesh is marketed as the answer to reliability issues of a wireless link. It certainly is a solution, but at what appears frequently as a ridiculous cost. The myth of mesh tends to be promoted by software companies who lack the RF knowledge to realise that better system design primarily resides in a robust radio. Instead, we see standards that resemble an inverted iceberg, with minimal radio technology at the base, and a massive software overhead which attempts to compensate for the basic failings.

The other myth of mesh is its association with low power. While a sensor node within a mesh can perform as a low power node, as soon as it become involved with routing within the mesh, the power and memory requirements start to spiral. These devices, which form the backbone of the mesh, need power. Inevitably such meshes become complex to configure, resulting in implementations that require careful site surveys and lengthy configuration.

Wibree brings the opportunity of debunking the myth. Its robust radio removes the requirement of building a mesh simply to bring reliability to a radio network, as that reliability has been designed as core technology. Where there is a need to extend range beyond normal operating limits, it becomes far cheaper to deploy additional access points and relay nodes. Like mesh nodes, these will need to be powered, but the incorporation of the attribute protocol makes them much easier to set up, allowing simpler reconfiguration and extension of the network."

His full article is here



The Industrial Ethernet Book - Articles: Wibree: a direct line to Industrial Wireless

Wibree, also know as Ultra Low Power Bluetooth, is poised to become the fastest growing wireless standard ever. Its symbiotic relationship with Bluetooth will open up substantial opportunities for network operators in consumer-based services. Less publicised are its many features that will prove of equal significance to the industrial market.
By Nick Hunn

When the Wibree standard was announced by Nokia in November 2006, it caught much of the industry unawares. The initial response from many analysts was simply to categorise it as yet another competitor in the 2.4GHz space. Others pronounced it to be just a ‘Bluetooth spoiler. Nothing could be further from the truth. In 2007, Wibree was adopted by the Bluetooth SIG as the ultra low power arm of the Bluetooth family. Indeed the concept envisages dual-mode chips that can support both Bluetooth and Wibree. This symbiotic existence will be key to Wibree’s market success. There will also be single-mode low power chips which enable a wide range of devices to talk to these dual mode chips.

Every wireless standard aimed at mass market applications faces the problem of achieving a critical mass of nodes. Wi-Fi managed this on the back of laptops, Bluetooth managed it on the back of mobile phones. As yet none of the other prospective short range wireless technologies have found a platform that will generate critical mass within their market place. This is where the Wibree design scores: it has a built-in route to mass deployment. Because mobile phones containing Bluetooth chips shipped in Christmas 2008 will include Wibree dual-mode functionality – effectively for free – it means that by the end of 2009 there could be over 100 million Wibree-enabled handsets in existence.

From the first to the millionth 802.11 chip took a leisurely four years. Bluetooth did better, but was still a slow starter taking 17 months from the first product to the millionth one, although it proved exceptionally active since that point, taking just another five years to get to the billion mark. Because Wibree will be automatically included within new generation Bluetooth chips, it is likely to reach that one million shipment milestone in just a week ...

That concept of Wibree within a Bluetooth chip is vitally important in understanding its place and the role that it can fulfil. Because low power, personal Wibree devices will be able to communicate with handsets, it means that in time every mobile phone becomes a Wibree gateway to the mobile network. So every Wibree device can communicate with the internet, allowing information to be sent backwards and forwards. And because the data rates are low, the cost of this data transfer will be a negligible portion of the user’s monthly phone contract. The new paradigm will enable a range of additional services that today are just too expensive for widespread deployment.

Where Wibree came from
Wibree didn’t just appear from out of the blue. There is an irony in the fact that the origins of Wibree appeared as the alternative proposal for the radio and Media Access Controller (MAC) for the IEEE 802.15.4 standard, which is now the basis of ZigBee and other short range radio networks.

Back in 2001 two industry groups put forward proposals. Nokia headed one of the groups and proposed a handset-centric development. This group’s major design tenet was that ‘it could be deployed with minor effort into devices already having Bluetooth, e.g. cellphones’ with the added requirement that a ‘common RF section with Bluetooth must be possible’. This vision was also broader than that of the competing camp; it envisaged a world of a trillion wireless, web connected devices. A key slide shows millions of connected laptops, billions of mobile phones and trillions of what could be interpreted as Wibree-enabled devices.

In the event, the IEEE committee chose to adopt the alternative proposal for the 802.15.4 standard – but this did not stop Nokia working on its proposal. It matured in a commercial project supported by the Finnish Government.

The key architect of the Bluetooth standard continues to add its expertise for what will be the most robust and effective short range wireless standard yet. Those interested in watching a standard evolve can still see the original proposals on the IEEE site1 and compare them as the final version is released.

What Wibree does
Wibree provides ultra low power radio within the 2.4GHz band. Low power is always determined in large part by the application: the longer a device is active and the more data it transmits, the shorter will be its battery life. The declared aim is a radio that can transmit a small packet of data approximately every second for a year using a small button cell such as a 150mAH CR2032. If the duty cycle is reduced to one transmission every 15 to 30 seconds, then the battery life effectively becomes the leakage life of the battery. A key radio design consideration limits peak current to less than 15mA – the highest current a button cell can provide – while achieving a range exceeding 100 metres. This is beyond the bounds of any current radio technology of significant range, including Bluetooth and ZigBee.

The overall low power drain derives from a protocol that lets the radio stay asleep for most of its life. It can wake up quickly where upon it broadcasts its requirement to transfer data on a number of advertising channels across the spectrum. The receiving device, which is likely to contain a larger battery as it will be on for more of the time, acknowledges receipt of this data, at which point both can go back to sleep. The whole process will take less than a few milliseconds.

Wibree within an existing Bluetooth chipset represents a key cost advantage, but not the only one coming from a symbiotic existence. A growing level of interference has to be a major concern about radio deployment in the 2.4GHz band. Bluetooth for industrial applications already exploits this concern through its resilience to interference. Where ultra low power is a requirement, there is still no satisfactory solution – a situation that has persuaded groups such as ISA2 to look afresh at their radio requirements for a robust industrial wireless standard.


Wibree technology protocol stack:
Wibree was designed with two implementation alternatives. The stand-alone implementation is designed for use with applications which require extremely low power consumption, small size and low cost. Examples might include watches, sports and medical devices and human interface devices (HID) such as wireless keyboards.

The Bluetooth-Wibree dual-mode implementation has Wibree functionality integrated with Bluetooth by reusing key Bluetooth components and the existing Bluetooth RF. Examples of devices that might benefit from the dual-mode implementation include mobile phones and personal computers. The Wibree link layer provides ultra low power idle mode operation, simple device discovery and reliable point-to-multipoint data transfer with advanced power-save and encryption functionalities .The link layer provides the means to schedule Wibree traffic in between Bluetooth transmissions. echo route.

Wibree provides the answer. As with the Adaptive Frequency Hopping of Bluetooth v1.2, the same techniques have been introduced into the Wibree standard, but in a form that removes the need for Bluetooth’s high accuracy crystals. This preserves the dual objectives of low power and low cost that characterise Wibree.

This robustness is critically important for industrial applications. The 2.4GHz band is unlicensed and widely used by many different radios, including Wi-Fi and ZigBee. Over the course of an industrial project, it’s difficult to anticipate what level of interference will be encountered over the life of the installation. Direct Sequence radios that sit on a single channel (such as Wi-Fi and ZigBee) are notoriously prone to interference, to the extent that they can stop operating. Adaptive Frequency Hopping can adjust the frequency usage to follow a changing spectral picture. This characteristic alone promotes it the method of choice for EMC-toughened industrial applications.

In many cases, the other end of the Wibree link may be a mobile phone acting as a gateway or alternatively a Wi-Fi access point. The processor running these can accumulate the frequency pattern that the Bluetooth chip performs, along with knowledge of the Wi-Fi channels to ‘pre-program’ the Wibree hopping scheme to ensure best performance, even where competing radios are co-located.

Questions about range
Much of the early coverage of Wibree concentrated on its use in short range applications, typically for sports and fashion devices that would connect to a mobile phone. The working groups that are evolving the standard understand the importance of the wider Machine to Machine (M2M) market and are ensuring that Wibree has the capability to address them. This short range label is the same type of understatement that has haunted Bluetooth; although Bluetooth is normally referred to as a short range technology for less than ten metres, the reality is that it is successfully used for many applications over hundreds of metres3.

Looking more closely at the parameters that will determine range, the first point is that it will share the Bluetooth radio and receiver chips. The most recent generation devices have receive sensitivities around - 85dBm and can directly deliver about +4dBm of transmit power. With careful RF design this gives an open field range better than 200 metres. The higher modulation index of Wibree suggests that for the same receive and transmit values, the link budget should offer an additional 20% range increment. Dual mode Wibree chips will use the same receiver and transmitter technology within these chips for sensors typically within the house or factory floor. Adding a power amplifier to boost the output to 100mW (+20dBm) should make it possible to achieve an open field range close to a kilometre. That will require the addition of power control and a possible wait for a future release of the standard. But as more designers see the potential, it is certain that it will be demanded.

The initial profile set
Wibree is adopting the principle of profiles to define its most common application areas. In its initial release, these will cover intelligent displays, such as a watch and sensors, where the standard focuses on industrial and medical applications. Although this may seem a somewhat esoteric selection, the standards behind them enable far more than a first glance would suggest.


This is not Wibree yet, but indicative of hardware modules to be available in 2008.The device shown here is a fully integrated and qualified Class 1 Bluetooth module with RF, baseband, antenna and protocol stack. It can deliver up to 65mW of RF power giving line of sight operation up to a kilometre. This module uses the CSR BlueCore 04 chipset. [EZURIO]

The Generic User Interface Device profile provides a method of transmitting information to any display. And the most prevalent portable display is the screen of a mobile phone. Turning this around, the Generic User Interface Device profile may be employed to make a handset a general purpose display for other devices. That can be anywhere. At home, or in the wider world, such as public transport information broadcast from a bus stop or in a railway carriage.

The receiving device doesn’t need to be static for this scenario. A feature of the short time required to complete a data transfer means this profile can be used with moving receivers. If we consider a transmitter with a 100m range, a vehicle moving at 100km/h will be within range of the transmitter for around four seconds – more than enough time to pick up traffic information from a beacon. An increasing number of vehicles already have a Bluetooth-enabled driver display – aka satellite navigation system. There is minimal incremental cost using Wibree to receive additional messages from roadside transmitters. This lends itself to Integrated Traffic System application development.

In the industrial environment, wireless sensing is potentially an enormous market. It doesn’t just cover industrial SCADA in factories, it encompasses active ‘pulling’ of information from any device that needs to send information. The low power of Wibree enables a host of battery powered devices. It also opens up the market for ‘power-free’ devices that either use solar, thermal or vibrational energy conversion sources.

Network architecture
An important facet of the Wibree standard relates to the work being done to connect sensor devices to a remote server. Most wireless standards limit their scope to the immediate wireless connection. Wibree contains a new Attribute Protocol, which views overall platform development end-to-end. It allows a Wibree sensor to locate a gateway device and connect to it, then use its attribute protocol to extend its connection through the gateway to a predetermined remote middleware location. From this point the attribute protocol lets the sensor converse directly with the middleware, whether for control and interrogation, or output delivery to a remote database. The effect of this facility cannot be underemphasised. For the first time a wireless standard is being engineered that acknowledges the key network aspects of end-to-end platforms. It moves M2M from a specialist area to the possibility of low cost, mass deployments.

Putting these together, the Generic User Interface Device Profile, Sensor Profile and Attribute Protocol developments take in most of the prospective applications currently envisaged by wireless device developers.

Mobile network opportunities
Wibree embedded mobile phones turns these ubiquitous devices into gateways, in effect an integrated GPRS modem connecting to Wibreeenabled ultra low power sensors. The scope of these is limited only by demand and developer imagination. The obvious ones are healthcare. Less obvious ones will rapidly evolve. For example, consider emergency messages. If Wibree is fitted into the airbag in your car, then whenever it is deployed in a accident, the airbag could send an emergency call out through your phone. The cost of implementing that is around a dollar for the chip, plus the cost of the monitoring contract, which a network could offer for a minimal annual premium. Compare that to the cost of current systems, which involve several hundred dollars of hardware in the vehicle and a similar annual monitoring cost. It also plays to the current legislative requirements for mobile phones to provide emergency location information.

The same economics come to play in almost every scenario where a low cost alarm or monitor will be within range of a consumer handset.

The industrial market
In many ways, the industrial market has always had the greatest reason for implementing wireless. Unfortunately, the diversity of its applications has meant that it has had relatively little serious attention from those developing wireless standards. Even where a standard has presented itself as appropriate, it invariably has been designed with other priorities in mind that those of the industrial user.

Generally the prime requirement is robustness. Industrial applications need communications to work reliably and predictably. A second frequent requirement is low power consumption. Low power becomes important for several different reasons. It may be to remove the need for a power cable, whether because of the cost of a power connection, or because the sensor is located on a moving part where power cannot be provided. If both are to be achieved the best approach is to design them into the core standard as an integral part of its architecture, rather than trying to bolt them on afterwards.

In reality, this has not been addressed by the majority of wireless standards. Specific applications have successfully used proprietary radio design, while standards-based activities have mostly concentrated on higher volume consumer markets, either ignoring, or failing to evangelise their industrial potential. As a result, a number of unfortunate myths continue to circulate.


The ultimate industrial HMI? It is really not that so far fetched with the inclusion of Wibree ultra low power Bluetooth silicon [CSR]

One of the most pernicious myths relates to mesh networks. Meshes have many admirable qualities when they are appropriately used. However, in many cases mesh is marketed as the answer to reliability issues of a wireless link. It certainly is a solution, but at what appears frequently as a ridiculous cost. The myth of mesh tends to be promoted by software companies who lack the RF knowledge to realise that better system design primarily resides in a robust radio. Instead, we see standards that resemble an inverted iceberg, with minimal radio technology at the base, and a massive software overhead which attempts to compensate for the basic failings.

The other myth of mesh is its association with low power. While a sensor node within a mesh can perform as a low power node, as soon as it become involved with routing within the mesh, the power and memory requirements start to spiral. These devices, which form the backbone of the mesh, need power. Inevitably such meshes become complex to configure, resulting in implementations that require careful site surveys and lengthy configuration.

Wibree brings the opportunity of debunking the myth. Its robust radio removes the requirement of building a mesh simply to bring reliability to a radio network, as that reliability has been designed as core technology. Where there is a need to extend range beyond normal operating limits, it becomes far cheaper to deploy additional access points and relay nodes. Like mesh nodes, these will need to be powered, but the incorporation of the attribute protocol makes them much easier to set up, allowing simpler reconfiguration and extension of the network.

It must be emphasised that neither Bluetooth, nor Wibree are designed for high data rate applications. They typically provide throughputs in the hundred of kilobits per second. If low power is paramount, then the amount of data being transferred needs to tumble with the device itself being powered down for the majority of its life. Where industrial applications need sustained higher data rates, either Bluetooth EDR (Enhanced Data Rate) or 802.11b/g (Wi-Fi) is likely to be more appropriate.

Time to start planning
Bluetooth, Wi-Fi and other short range wireless technologies have already spawned a first generation of personal wireless devices and embedded applications. However, neither has made great inroads in addressing the dual requirements of the myriad of devices that must operate off a small battery for a period of years and also have a readily available portal to send their information back to the internet. Nor have they successfully penetrated a significant portion of the industrial market. Wibree ticks the boxes to change this.

More importantly, Wibree comes built within Bluetooth chips. Current development in Bluetooth, with the evolution of a medical profile and industrial strength automation profiles are paralleling the same developments within Wibree. A product designer can start to design with Bluetooth today, knowing that there is a low power roadmap to Wibree within a few years.

Taking all of these factors together, Wibree has the potential to transform SCADA and Process applications. By making the mobile handset a gateway, it brings the network operators into the equation. And they have the resources to aggregate and enable service provision.

With Wibree adopted within the Bluetooth fold, more resources and expertise are being stitched into the fabric of the standard. However, it is actively supported by the major Bluetooth chip vendors including Cambridge Silicon Radio and Broadcom. Its presence will hinder other short range, low power wireless technologies from gaining traction in the handset, ensuring that Wibree is placed to own the low power wireless market, whether in industrial applications, healthcare or consumer products.

We will shortly see the first modular Wibree platform development kits. In the meantime expect the mobile networks to engage in leveraging their infrastructure for short range wireless applications.

References
1. http://www.ieee802.org/15/pub/2001/Jul01/ Original proposals to the IEEE 802.15.4 working group. Documents reflecting the genesis of Wibree include: 01230r1P802-15_TG4-Nokia-MAC-Proposal1.ppt 01231r1P802-15_TG4-Nokia-PHY-Proposal1.ppt
2. http://www.isa.org/MSTemplate.cfm?MicrositeID=1134&CommitteeID=681ISA Working group SP100 – Wireless Systems for Automation
3. http://www.ezurio.com/dl/open/?id=19Understanding Range – a white paper explaining wireless range

Nick Hunn is CTO, Ezurio Ltd

No comments:

Post a Comment