The Bluetooth Invasion Begins

Bluetooth, a new wireless interconnect technology standard, is designed to replace the many cables we need to connect devices to our PCs or networks. Called a PAN (personal area network), this technology will let you walk into a room and communicate with your PC or other Bluetooth-compliant devices. The Bluetooth specification -- whose name is derived from an aggressive Viking king so fond of blueberries that they stained his teeth -- was created in February 1998 by a consortium comprising inventive minds from Ericsson, IBM, Intel, Nokia, and Toshiba.

Support for Bluetooth is solid, with more than 1,880 adopter/associate member companies signed up to the Bluetooth SIG. Although the technology has been the subject of tremendous hype (some reports would have us using Bluetooth in everything from refrigerators to toilets) and has the potential to explode on the market, Bluetooth was not designed for full WAN connectivity. At best, the technology can be used for ad hoc networking when needed, but it is not designed for more than limited use. While other standards, like 802.11b, are designed for wide-reaching applications, Bluetooth simply does not have the bandwidth to handle full network demands.

Bluetooth is a good, low-power data/voice-transmission standard that will be a real boon for certain applications. Imagine a cell phone or pager that automatically transfers to vibrate mode when you enter a restaurant or theater. Or a laptop and GPS (global positioning system) that automatically feed a display in your car for directions. One of Bluetooth's design strengths is its low cost. The Bluetooth SIG says integrating this technology should add only $5 to the cost of a unit.

Specifications

Bluetooth operates in the 2.4-GHz ISM (industrial, scientific, and medical) band and uses a fast frequency-hopping technology to minimize interference to and from non-Bluetooth sources. This frequency-hopping occurs nominally at 1,600 hops per second. The system has 78 possible channels. Channel spacing is 1 MHz in every industrialized country but France. (In France, you are limited to 22 possible channels on which to hop. Negotiations have been ongoing with the French government to change its regulations.)

Bluetooth has three power classes for transmitting: 100 milliwatt, 2.5 milliwatt, and 1 milliwatt. The range for 100 milliwatt is 100 meters, 2.5 milliwatt is 10 meters, and 1 milliwatt is 10 centimeters. Remember that these ranges are variable and difficult to calculate. Materials, walls, and interference from other 2.4-GHz sources all can change the range achieved. Digianswer (a division of Motorola), a DSP and Bluetooth developer, has proven that a 2.5-milliwatt PC Card can reach up to 100 meters in a completely unobstructed environment.

Throughput for Bluetooth is rated at 1 Mbps under optimal conditions. With error-correction overhead and other environmental factors, however, the real throughput should be about 700 Kbps to 800 Kbps. This should be more than enough for the Bluetooth cable-replacement designation.

Several other factors can influence Bluetooth's speed. The presence of other 2.4-GHz devices, such as an 802.11b WAN, can degrade performance of a Bluetooth piconet. Bluetooth traffic is separated into two types: data and voice.

Voice traffic has a much lower tolerance for interference than does data traffic. Ericsson conducted a study of interference from 802.11b on Bluetooth and found that when 802.11b is operating at normal traffic conditions, the more sensitive voice operation of Bluetooth is not affected when the transmitter and receiver are less than two meters apart. At a range of 10 meters, the probability of noticeable interference increases to 8 percent.

The data link gets more interference than the voice link, but is less susceptible to it. At a 10-meter range, throughput loss of more than 10 percent can occur with a 24 percent probability. The Ericsson study reports that the total loss of throughput due to interference from 802.11 WANs cannot exceed 22 percent. The 801.11 networks take up 17 channels at 2.4 GHz, and Bluetooth has 79 available channels. As of press time, I was unable to get any data on Bluetooth's effect on 802.11's throughput. The IEEE's 802.15 working group is developing standards to let Bluetooth and 802.11b function better in shared space (see grouper.ieee.org/groups/802/15/).

Factor in Microwaves

The FHSS (frequency-hopping spread spectrum) protocol that Bluetooth uses is sensitive to interference from microwave ovens. Basically, a frequency-hopping protocol like FHSS cannot avoid channels that have been obliterated by noise. It must hop through all 79 channels on a continuous cycle. Packet and frame loss will occur when it jumps into a noisy channel.

Security

Security for Bluetooth devices is less than wonderful. It provides link-level security plus encryption. Security in a Bluetooth network is entirely device-based, not user-based, as in traditional systems. Bluetooth has three levels of security:

* Level 1: No security.

* Level 2: Service-level-enforced security. Security is established after channel negotiation.

* Level 3: Link-level-enforced security.

The Bluetooth SIG recommends that Level 2 be used in most instances (see www.bluetooth.com/developer/whitepaper/whitepaper.asp). If you go with Level 3 and full encryption, the ease of use and simplicity aspects of Bluetooth begin to fade. Level 3 would require user intervention for all services.

In Bluetooth, most security issues are expected to occur above the link layer, in the application or protocol layers. However, problems can occur at a purely physical level. It is also possible to cause DoS (denial of service) attacks against a wireless network by flooding the 2.4-GHz band with interference. Wireless networks are also vulnerable to passive eavesdropping attacks.

In this scenario, a potential hacker could simply listen for Bluetooth packets and extract data from them. The frequency-hopping characteristic of Bluetooth largely eliminates this problem. The hacker would have to know the exact sequence of hops and channels at 1,600 hops per second. We recommend using an external encryption program to pre-encrypt the data before sending it across Bluetooth.

Other wireless solutions use more robust solutions like WEP (Wired Equivalent Privacy). However, implementing something like WEP on top of Bluetooth would once again restrict simplicity and ease of use. It is important to keep in mind Bluetooth's design intentions when considering security in the standard.

Radios in Bluetooth can be master or slave, or be in simultaneous scenarios. Two possible Bluetooth networks exist: piconet and scatternet. When you bring Bluetooth radios within range of each other, they connect and form a piconet. One unit becomes a master, the other a slave. The master controls all the traffic in a piconet. Bluetooth radios in a piconet frequency-hop together. Each piconet can have up to seven simultaneous or more than 200 active slaves.

Radios in a piconet can be in one of five states: standby, inquire, page, connect, and park/hold. Standby is a radio waiting to join a piconet. Inquire is a radio seeking other radios to connect. Page is a master radio asking to connect to a specific radio. Connect is a radio active on a piconet as a master, slave, or simultaneous. Park/hold is a low-power connected state. The master gives all the slaves in a piconet its clock-device ID and sets the unique hopping sequence based on the master's device address.

Scatternets occur when multiple masters exist in range of each other. A master radio may also be a slave radio on another piconet. Each piconet is hopping with a different sequence sharing the same 2.4-GHz band. Because of the different hopping sequences, there is very little chance that any master will hit a channel at the same time as another master.

Bluetooth cannot hand off a slave to another master unit. This is a problem if you are sending voice data to a device and switch piconets. A couple of seconds of voice or even the connection could be lost. The specification offers no solution, because of the cost and complexity involved. Bluetooth is not meant to be a replacement for wireless LAN networks, but a cable replacement and ad hoc network.

Interoperability High Points

It would be insane for a vendor to put out a Bluetooth radio device without proper interoperability. The Bluetooth SIG conducts certification and keeps a list of qualified products on its website. The Bluetooth SIG and website (www.bluetooth.com) give fledgling Bluetooth developers all the help they need to create a compliant product.

When thinking about deploying Bluetooth, you should consider several issues. Bluetooth is suitable for everything from SOHO to full-scale enterprise use. Are you looking for a cable-replacement technology with phone and PDA benefits? Or are you looking for wireless networking?

Do you need a Bluetooth environment? Right now, no. Will you need a Bluetooth environment in the next two years? It's a real possibility. Company cell phones, PDAs, and laptops will likely be equipped with this new technology. Your end-user customers will ask for it, and executives will demand it.

When you deploy your Bluetooth environment, the first thing you have to determine is whether to make it a buildingwide deployment, a conference-room deployment, or just a couple of desktops. Buildingwide deployments give you the advantage of Bluetooth everywhere. If you're going to deploy buildingwide and have a wireless LAN or are planning to deploy one, consider the interference that these two technologies could give each other.

Conference-room deployment gives you ad hoc file sharing and other benefits. However, conference-room deployments could also react with a wireless LAN environment, but on a much smaller scale.

The advantage of personal deployment is realized only if your Bluetooth commitment is very small. Personal deployments should have almost no effect on the wireless LAN environment. If you are going to do anything on a conference-room scale or beyond, see what it would take to deploy this technology to the entire enterprise and design your implementations to take advantage of the conference-room equipment you are going to install.

Bluetooth will come through on its promise of interoperability in the next two years. This quarter will offer a few new products and a better perspective on how Bluetooth is going to affect us. Motorola's Timeport 270 cell phone and the GN Netcom GN9000 Bluetooth headset are now using Bluetooth.

As the technology matures, the cost of Bluetooth devices will fall. Right now, Toshiba offers a PC Card Bluetooth interface, called the PA3053U. These cards list for $169 on the company's website. I put one of the Toshiba cards in a Compaq Computer Corp. 1600 laptop and the other in a Dell Computer Corp. Latitude laptop. I loaded a few drivers, rebooted, and indeed experienced ad hoc file and print sharing just as advertised. The number of devices will increase quite a bit in the next six months. Be ready for the invasion.