For those of us who have been involved in technology development for wireless networking, the medium access control (MAC) layer is fairly well understood. But as wireless local-area network (WLAN) technology rolls out to wider audiences, who must implement wireless capability at their company, there are sometimes misconceptions about MACs. I've even heard some newcomers to the field say that it doesn't apply to them because they don't have Mac computers in their network!
So, for those who aren't real sure what MACs are, I offer a description of this function, a look ahead at future development and also considerations for selection of an appropriate MAC.
First, what the heck is a MAC?
MAC stands for "Medium Access Control." The term can apply to a single IC, or to the MAC "layer" in a wireless system. In a WLAN, the first layer is the physical layer (PHY) that handles data transmission via some wireless media (RF or infrared, for instance). The MAC layer comes next and serves as the interface between the PHY layer and the host device.
So, you'll find that when the MAC is referenced by a WLAN radio designer, they're probably talking about the MAC layer. The MAC layer is resident in or on every WLAN adapter card and Access Point. This layer is the central processor that formats data bits into packets, and manages the transmission of that packet information over an RF medium to guarantee successful connections between one WLAN radio and the next.
The MAC provides four key functions. First, it provides the basic I/O control, configuration, and frequency channel selection of the physical layer (the radio). Secondly, it serves as the host interface, typically supporting PC Card, PCI, or ISA standard interfaces between the radio and the PC platform. Thirdly, the MAC provides the necessary network management functions such as authentication and other security methods such as encryption, which are necessary to assure the security of information transmitted over a wireless LAN network. And lastly, the MAC is used to translate data packets between wired Ethernet LANs and WLAN systems.
In the wired LAN world, the function that a MAC provides is handled by an Ethernet controller card. Essentially, a MAC in a WLAN does the same thing as an Ethernet controller. Those controllers are generally hardware based because the standards for them are mature and interoperable, having been in place for about 10 years.
WLANs, on the other hand, have been implemented around software/firmware-based or RISC-core-based MACs, because of the infancy of both markets and governing standards. Before the creation of the IEEE 802.11 standard, many of the MACs employed were Intel microprocessors (i.e. 186, 188, 286). The use of these processors limited data rates to about 2 megabits per second. This is because of the high processing overhead necessary for the software/firmware to implement the protocol.
The present and future
As demand for much higher data rates are felt in several wireless market segments, and as various standards/specs emerge for those spaces (i.e. 802.11, Bluetooth, HRFWG, HiperLAN, etc.), it's apparent that MACs will have to achieve faster data rates, plus offer special benefits for given protocols and applications. Even now, the 802.11 standard that was approved just a year ago is finalizing the draft for an extension to 11 Mb/s.
It's easy to conceptualize data rates going even higher before jumping to another frequency. Just look at the home-networking segment. Initially, members of the WLAN industry assumed that the consumer and home segments would not require the speed of enterprise segments. That notion has been entirely supplanted by the increasing demand in the home for full-up, multimedia content. That certainly won't go away when the home network goes wireless. All of these developments will force MAC designers to rethink their architectures.
A RISCy business
I think there is great promise in migrating MAC technology to RISC-like IC machines, especially for implementing protocols bounded by time critical specifications. It's been proven that RISC architectures using today's sub-micron CMOS semiconductor technology can achieve data rates well above 25 Mb/s.
RISC architectures have optimized firmware instructions for implementing complex protocols at a fraction of the processor's bandwidth over conventional microprocessors. This is really attractive to WLAN system programmers. In addition, as we look ahead to further integration of functions to bring costs in line with critical consumer price points, the use of RISC architectures opens the door for integration of such a MAC with the baseband processor.
A RISC-based MAC is also flexible in it's ability to support several protocols instead of being slave to just one, and that flexibility is achieved without sacrificing speed and throughput.
Things to consider
There are number of issues to consider when selecting MAC technology for your next WLAN solution. First, consider what software drivers and firmware tools you'll get with the MAC you select. A good suite of software and firmware tools is critical when developing the MAC protocol for a standards-based WLAN application. This is true because MAC software engineers usually spend a good portion of their time understanding the specifics of a particular standard's protocol in parallel with the development of firmware.
The second issue involves host interfaces. It's equally important to plan ahead and determine your target host interface so that you do not have to add extra glue logic ICs between the MAC layer and host platform. Having to add external glue interface devices is a disadvantage, in that it adds to the bill of materials cost. So choose the host interface firstwhether it's PC Card bus, ISA, USB, PCI or Mini-PCIthen, choose the MAC that best serves that interface.
The third, and maybe most important of all considerations, is the choice of MAC technology that provides a path to higher data rates. Let's face it, once you commit to a given architecture, you're in it for the long haul. You'll most likely develop a suite of firmware to satisfy a number of protocols and you'll want to re-use and capitalize on your investment. Therefore, it's important to choose a MAC layer platform that allows upward portability of software and firmware at a minimal investment.
It is my experience that as high rate standards emerge, changes will affect primarily the PHY, and not necessarily the MAC layer. And even if there are changes affecting the MAC layer, you'll want to make sure the processing power is available to accommodate such modifications.
When you make a commitment to a given architecture, ensure that your selection will last through a three-year life cycle. The odds are you're going to invest in the development of firmware and software for a variety of protocols, so make sure that the RISC architecture (or processing platform) selected will serve the target markets.
In closing, MAC technology has come a long way since the first Ethernet controller. As networking goes wireless, standards continue to emerge and markets demand higher and higher data rates, selection of the right MAC will become more and more critical to system success. So take your MAC seriouslythoroughly research standards and migration paths, and make sure you've picked a MAC technology that will serve your system well down the road.
Look for more about MAC technologies to deliver voice and data traffic in upcoming columns. If there are other topics you'd like covered, please let me know. ‘Til next time, remember that I really enjoy hearing from you, so keep those emails coming in.
About the Author:
Al Petrick is the Senior Manager of Strategic Marketing for Wireless Products at Harris Semiconductor. He is also the Vice Chairman of the IEEE 802.11 standards committee. He holds a BSET degree from the Rochester Institute of Technology (Rochester, NY) and an MBA from the Crummer School of Graduate Business at Rollins College (Winter Park, FL). Al can be reached at firstname.lastname@example.org.