March 07, 2006

Making USB without wires work for consumers

There are at least three flavors of wireless USB: Cable-Free USB, Certified-Wireless USB and USB-to-TCP/IP converter. This article evaluates each and concludes Cable-Free USB is best for ultrawideband without wires.

The Universal Serial Bus (USB) is one of today's most useful and prevalent connection methods. So many electronic components in the market—from cameras to computers to cell phones—have USB-enabled connections that hubs were created to allow even more devices to be connected to a computer than the computer has physical connections. All those connections are accompanied by plenty of wires, each representing a limitation on the consumer.

There are several physical connector types for USB and since physical connection must matched between devices, there's plenty of potential for confusion. As USB devices become more visible in homes and offices, it’s desirable to reduce the wiring to allow better device placement and reduce clutter and complexity.

Wireless connections are the obvious solution. With so many devices coming from different markets and a wide variety of companies, there are different concepts for solving the problem of getting rid of the USB cable.

There are three basic strategies for transferring USB signals wirelessly:

  1. The USB serial data stream is converted directly into Ultra-Wideband (UWB) radio signals. The devices do not need to know if the USB is transported through a cable or through a wireless signal.
  2. Use only the USB connector, but repackage the data so that the connector is really sending something else, usually TCP/IP.
  3. Redefine USB internally to electronic systems, and only preserve compatibility at the applications layer of software. This is the least backward-compatible solution.

There are several solutions for wireless USB on the market today, and emerging solutions that are expected in 2006 and beyond.

Strategy #1: Cable-Free USB
Cable-Free USB, a combination of Freescale Semiconductor’s Direct Sequence Ultra-Wideband (DS-UWB) chipset and Icron Technologies’ ExtremeUSB—Wireless Edition technology, enables a true USB connection wirelessly.

This technology was demonstrated in products at CES 2006. Cable-Free USB is compatible with the USB 2.0 standard and fully backward-compatible with previous USB standards. No driver changes or software installation are required; the technology supports the plug-and-play functionality defined by the USB Implementers Forum. With Enhanced Host Controller Interface (EHCI) and Universal Host Controller Interface (UHCI) support, no host controller changes are required. These criteria make Cable-Free USB a very user-friendly solution.

Cable-Free USB supports major operating systems that support USB (including Microsoft Windows 2000, Windows XP, MAC OS X, Solaris, Unix and Linux) without modifications to drivers or the OS. It also supports USB transfer types, including control, bulk, interrupt and isochronous.

Security is the same as that of wired USB; it is provided at the UWB PHY/MAC layer using Advanced Encryption Standard (AES) 128-bit encryption with RSA key exchange. The technology supports many association models including factory matched pairs, one time push button association and near-field communication (NFC). Cable-Free USB provides an equivalent experience to wired USB, including support of up to 127 connections.

DS-UWB radio technology allows very high data rates at very low transmission power, with the added benefit of requiring very little system power consumption. Freescale’s rule of thumb for power consumption is one megabit of transmission rate for one milliWatt of system consumption.

Cable-Free USB is a "true USB" interface to the high-performance DS-UWB radio system which strictly adheres to the USB 2.0 interface. There are several advantages to strictly adhering to the USB standard. A DS-UWB-enabled system can connect to a USB-enabled system without any software updates required, because it is based on the USB 2.0 standard.

A DS-UWB-enabled system can be placed on a mini-card form factor and use USB connections directly, without modification to the host system. This approach can help connect the more than 700 million USB devices in the market—as they exist today, with no changes—and provides a solution today for achieving USB wireless operation. Figure 1 illustrates some wireless USB applications.

Click here for Figure 1

1. Connection scheme for Cable-Free USB enabled hub: How USB and UWB are combined.

Using USB in an RF environment introduces inherent difficulties of delay, latency and power management.

Originally, Icron’s ExtremeUSB technology was developed to enable USB devices to be used in industrial and commercial environments where operational requirements (typically the 5 meter limit) often exceed those of the desktop for which USB was designed. In particular, this technology overcomes the limitations imposed by the turnaround timer (TT).

Removing this limitation enables the conventional RF techniques such as error correction to be used. Operating at the USB protocol layer, the ExtremeUSB stack is independent of the physical media used for data transmission.

The TT parameter limits the time that any host or device may take to respond to a request or to acknowledge data reception. The TT was introduced into USB design to prevent an errant device from consuming a disproportionate amount of time on the shared bus. It helps maintain high bus occupancy by limiting the time that the bus sits idle waiting for a response that may never arrive.

In wired USB, the TT budget is allocated to the various hubs, cables and devices that constitute the worst-case (greatest delay) topology.

In wireless USB, there are additional factors that can push the turnaround time beyond the allowed limits. Some radios take longer than others to switch between RF transmit and receive modes. Restricted RF bandwidth can force longer transmission times. Lost packets equate to an infinite TT. Adding error correction, scrambling and encryption functions increases latency and consumes additional bandwidth.

Another major difference between wired and wireless communications is the much higher error rate at the physical layer that must be expected. Compensating for this requires the use of error correction and scrambling functions that exacerbate the TT issue mentioned earlier. In the limiting case, when the error rate becomes very high, re-transmission of an entire packet may be required.

Standard USB was not designed with these issues in mind and does not accommodate them well, if at all. Cable-Free USB is designed to enable standard USB hosts, hubs and devices to be used in high-delay situations. In addition, any particular implementation can combine support for speed and transfer-type variants as required. Just like standard USB, devices with different speed and transfer type attributes can be attached to and detached from the system at random. Cable-Free USB recognizes each device automatically and provides the appropriate protocol handling.

Click here for Figure 2

2. Icron’s ExtremeUSB Stack.

As shown in Figure 2, the zero-install Extreme USB does not affect the systems that it connects to—no software installation is required. The Cable-Free USB solution is engineered to be self-contained.

Strategy #2: Repackage the data and reuse the USB connector
Another strategy is to repackage the UWB signal through the USB connectors. One solution that is being proposed in the industry does this with a dongle and remote USB hub.

On the host side (typically a PC), the dongle is attached. Software must be installed on the PC. This allows the PC to intercept the USB communications at the software level, then reformulate and send it as a TCP/IP signal through the USB connector, and then over the air. This, in turn, is received by the hub, which is really a TCP/IP-to-USB converter box.

This strategy provides for some backwards compatibility by using the existing USB connectors, but it does require driver updates and adding a TCP/IP-to-USB conversion layer.

A disadvantage of the TCP/IP solution is that isochronous communications are not possible, which means that systems engineers must build in extraordinary amounts of communications headroom—and that means extra expense—to achieve trouble-free performance for audiovisual systems.

While using existing USB connectors is very good for the market, ultimately the best market solution should not require a bulk software change at the driver level for thousands and thousands of device types and countless millions of lines of code.

Another important limitation of this solution is that because TCP/IP is being transported, this type of system can only be an interim solution on the way to interoperating with other Certified Wireless USB systems.

Products with this capability embedded, or even as dongles, will not work with the emerging Wireless USB standard and create a second "new flavor" of USB for consumers to juggle.

Strategy #3: Redefine USB for wireless
A third solution is to redefine the USB specification for wireless and account for the goal of allowing local interconnections of devices wirelessly. In essence, the only portion of the original USB that would remain is the software connections from the applications layer into the software driver layer; everything else would change—including hardware, drivers and even system architecture.

This type of system is USB almost in name only and in fact does go by a new name: Certified Wireless USB. This solution aims to use a different approach to DS-UWB, called Multiband Orthogonal Frequency Division Multiplexing (MB-OFDM).

An advantage to this type of approach is that systems can be designed for this specification and optimized for this solution. This is true with almost every ground-up system design. But the implications for an established USB ecosystem should not be underestimated.

So, what is a provider of systems that’s motivated by profit supposed to do today? Rather than design for the future and wait for sales to happen, most providers would be forced to have both a wired USB solution and a wireless USB solution, with appropriate complexity and solutions for the system to properly handle both. Or they might take a wait-and-see approach to the market.

For example, for mini-cards to be implemented on a PC, wired USB signals will be present since the specification allows both USB and the PCI Express interconnect technology to be used.

Therefore, most systems would not eliminate the USB bus anyway. Even though the USB connectors could be eliminated on PC systems that use wireless, the USB bus would remain and the software developers would still be required to write new software for this new wireless USB.

In other words, this method of implementing wireless USB is always an add-on to the system—not a substitute.

Finally, to underscore the significance of the software issue, it’s worth noting that most embedded systems, such as the systems in camcorders and phones, run operating systems other than Microsoft Windows. Therefore, redefining software drivers for USB means that you affect countless embedded software systems.

Obviously, as manufacturers evaluate their technology choices to enable a wireless USB solution, there are many issues to consider. Cable-Free USB enables these manufacturers to use the installed base of USB devices, enables "closed" systems such as existing cell phones, camcorders, printers, etc. to take advantage of wireless USB, and supports a variety of operating systems—anything and everything that USB devices support.

As Cable-Free USB technology becomes available to consumers throughout 2006, the demand for wireless freedom is expected to escalate. This may make the most important issue for manufacturers to consider: consumer happiness.

About the author
Calvin Harrison is manager of marketing and business development for Freescale Semiconductor’s UWB Operations.. He can be reached at calvin.harrison@freescale.com.