Q&A with Broadcom's Stephen Palm
Innovators 2008: The technical director of Broadcom's broadband-communication group discusses the progress and potential of 802.11n, powerline networking, and more.
By Brian Dipert, Senior Technical Editor -- EDN, June 26, 2008
|
Broadcom is strongly pushing 802.11n as a follow-on to the company's notable 802.11g success. Yet, as we've discussed on several occasions, 802.11n is still a "draft" specification from an IEEE perspective, (albeit one with some degree of interoperability secured, by virtue of the efforts of the Wi-Fi Alliance. Has 802.11n's lack of formal IEEE approval to date at all hampered the technology's adoption? Why has standardization taken so long, and when is it currently forecasted to occur?
The lack of a formal standard has not hampered 802.11n adoption. The availability of Wi-Fi-certified products is what matters most in today's marketplace, and the program has succeeded in driving consumer 802.11n purchases before the final standard. However, we expect that IEEE ratification will spur a wave of 802.11n deployments in enterprises, which are typically more cautious in their purchases than consumers.
The 802.11n standards process may seem to be taking longer than usual. This [delay] is primarily due to the sheer complexity of the specification, which offers an immense variety of optional modes and technologies for specific applications. [The IEEE included] several of the options for political reasons, and deployed systems will likely never implement them. An earlier hype cycle also exacerbated the seemingly long standardization process. Whereas most standards remain under wraps until they are nearly complete, all eyes have been on 802.11n since the very first steps in the process.
Should we expect to see any further notable spec tweaking prior to approval, and how will this impact existing 802.11n-inclusive equipment already in warehouses, on store shelves, and in customers' hands?
Given the stability of the baseline draft 2.0 specification, we do not expect any substantial changes that would lead to further delays in the standards process or affect already certified products. All of the major technical items in the specification have been resolved, and only relatively minor wording issues remain. We expect the final 802.11n standard to be ratified in the second half of 2009—a time period consistent with the IEEE process. All of Broadcom's draft-802.11n solutions the company has shipped to date should be able to support the final specification with a firmware upgrade. Currently, all Wi-Fi-certified products should continue to interoperate with Wi-Fi devices that are 802.11n certified.
I've noticed a diversity of 802.11n implementations in the market at diverse corresponding prices. Take routers, for example. Some are 2.4-GHz-only. Others claim to be dual-band, although you can't have both bands operating at the same time. And still others support simultaneous 2.4- and 5.8-GHz "beacons," albeit with a per-band SSID [service-set identifier]. Some support channel bonding, and others don't. Then there are the Category 5-cable links. Some routers offer GbE (gigabit-Ethernet) connections for wired-LAN clients, synergistic with the higher PHY (physical)-layer speeds of 802.11n versus earlier Wi-Fi technologies, whereas others continue to rely on 10/100-Mbps-only ports.
Plenty of other examples of this plethora of specification specifics also exist. On the one hand, a specification with built-in flexibility is a good thing, because it enables a variety of implementations with different bills-of-materials costs. However, if you take this concept too far, it can result in confusion and consequent reluctance to purchase, interoperability problems, and other issues. Can you address these problems?
Broadcom's dual-band 802.11n solutions can operate simultaneously in both bands, but manufacturers can also implement dual-band systems that operate consecutively. Although 802.11a has been available for many years and [manufacturers could have implemented dual-band products with 802.11a and 802.11g, this approach was not popular due to cost and lack of need. With the rise of interest in distributing video using Wi-Fi, there has been a significant increase of attention for 5-GHz usage for video while maintaining data/voice compatibility using 2.4 GHz. With the additional integration of 5-GHz radios, simultaneous dual-band functions will become increasing popular as the end-user price declines. This scalability is one of the benefits of 802.11n, and manufacturers can include the exact number of features and components to suit the application of their products and remain completely interoperable with other 802.11n products.
The 802.11n spec includes many optional features that make it applicable to a variety of devices and usage cases. This flexibility enables our customers to differentiate their offerings for specific applications with basic, mid-tier, and high-end products at various price points. Ultimately, we do not feel that having different mixes of 802.11n features will create purchase paralysis or interoperability issues. The Wi-Fi certification process will ensure that products work together, regardless of any optional features that vendors apply.
Today, Wi-Fi certification mandates baseline interoperability in a single band, meaning that a 2×2 router in the 2.4-GHz band will communicate with a 1×1, 3×3, or 4×4 client in the same band. In the future, certification may involve optional features for certain applications, such as handheld devices using single-stream 802.11n or consumer electronics with techniques making video more reliable. But we expect the Wi-Fi Alliance to keep pace as the demand for those features emerges.
Broadcom is implementing some of the optional features and verifying that these features do not impact our ability to be Wi-Fi-certified. One of these features is the use of 40-MHz channels to provide peak performance. Because wider channels can cause interference, all Wi-Fi-certified draft-n products must implement a "good-neighbor" protocol that continuously scans the environment to alleviate interference problems. Broadcom is taking additional steps to minimize the interference effects of 40 MHz, including using algorithms that allow routers to step back to 20-MHz channel usage if the algorithm detects other devices. Broadcom is also shipping products in default 20-MHz mode, which forces the user to knowingly reset the parameters for additional performance, and the company is recommending that customers limit 40-MHz-channel usage to the less crowded 5-GHz band, although our manufacturing partners make the actual implementation decisions.
With all that being said, it is important to separate product stability or implementation issues from the way a feature was designed in the standard. [You shouldn't view] interoperability certification as a substitute for a company's quality-control efforts.
You seem less than enthusiastic about powerline networking, either as a competitor to or as a companion 'backbone' technology of Wi-Fi. Are you still of this mindset? What are your thoughts about other LAN- technology candidates, such as MOCA (Multimedia Over Cable Alliance), in which Broadcom staked a claim last May via its acquisition of Octalica, for example, or good old Category 5 cable? How do you see UWB (ultrawideband) technology, which promises to both act as a short-range wireless USB platform and as a longer range means of beaming high-resolution video and corresponding high-quality surround audio around premises, competing with or collaborating with Wi-Fi? How does your answer vary, depending on whether we're talking about a consumer's home or a corporate environment and whether the implementation will be in North America or somewhere else?
Broadcom continues to keep an eye on powerline networking, but, to date, we have not found a compelling enough advantage that will give it any significant momentum in the marketplace. Two significant detriments that the powerline industry must address are the multiplicity of incompatible standards, including HomePlug/AV, HomePlug, HD-PLC [high-definition powerline communication], UPA [Universal Powerline Association], and P1901, and poor throughput coverage. A single HD-video stream is likely to be available in only about 60% to 80% of homes with the current powerline technologies. Moreover, there is little that can be done to easily remediate a home. Another issue is that powerline basically can use only a single portion of spectrum unlike other technologies, which can use multiple noninterfering channels of spectrum. If a breakthrough in these areas is achieved in the near future, powerline networking may be a consideration. Otherwise, other network-connectivity technologies will continue to outperform and outprice powerline technology.
MOCA adds real value as a backbone technology, and, considering its strong support among telephone, digital-cable, and satellite-service operators, it will have a very strong position in the digital-home-networking landscape. MOCA technology supports more than four HD streams in 97% of homes without remediation, and installing a simple filter provides full coverage. MOCA offers multiple channels that allow multiple noninterfering networks in an area and relocation to avoid existing services on the coax. With the integration of MOCA into telephone, cable, and satellite SOCs [systems on chips], MOCA technology hits the price point for wide deployment.
Similarly, 802.11n has tremendous momentum as a backbone technology. As vendors continue to optimize 802.11n for video over Wi-Fi, it will likely emerge as a convenient and cost-effective way to network the entertainment portion of the digital home and, of course, remains the choice for mobile devices.
There are, of course, geographical considerations involving differences in construction materials and home size, which creates some variability in which backbone is most practical. But between Wi-Fi, MOCA, and, occasionally, Category 5 cable, we feel the majority of homes throughout the world can be accommodated for a compelling digital-media experience.
As far as corporate deployments, we do not see any wavering from Category 5 and Wi-Fi. With multiple cubicles and workstations on the same circuit, existing powerline technology would take us back to the days of shared 10-Mbps Ethernet of the 1980s. The main benefit of MOCA, reusing the existing coax cable in a high percentage of homes, is not exploitable in the corporate environment because coax cabling is just not there.
In answering these questions, what Wi-Fi evolutions, along with, perhaps, more disruptive revolutions, beyond 802.11n do you envision?
Wi-Fi will continue to proliferate into more mainstream products, and vendors will continue to offer faster, more robust connections. But the next big evolution will be integration of multiple technologies, including Wi-Fi, Bluetooth, GPS [global-positioning system], cellular, and mobile TV, onto a single die. If done right, these combinations enable manufacturers to differentiate the capabilities of mobile-phone handsets, gaming devices, and portable music players—without a significant impact on cost, size, or power consumption. Success with combinations will not only hinge upon a silicon vendor's existing technology portfolio, but also require a leap to lower process geometries. Broadcom has already transitioned to a 65-nm CMOS process, which we felt is a crucial step to staying ahead of the combination evolution.
There continues to be discussion about features and technologies for optimizing point-to-point and mesh-wireless connections. Many of these proposals have tended to be incremental in nature so it will be interesting to see if their benefit outweighs the current momentum.
One area that seems compelling is the point-to-point transmission of uncompressed video signals. Current wireless technologies provide data rates that are appropriate for networking MPEG-2 or AVC (MPEG-4) compressed-video streams. There has been increased interest to provide a wireless means of transmitting the final link from a set-top box or home-theater receiver to a wall-mounted HD display to ease installation issues. There is consideration for the usage of 60-GHz spectrum for this application, but cost-effective technology is still years away. Hopefully, the concept will not have too much hype as what happened with 802.11n.
Wired broadband is increasingly not the only option available. Take, for example, a cellular-broadband router that shares a common EVDO [evolution-data-optimized] Revision A, UMTS [universal-mobile-telecommunication-system], or, in the future, LTE [long-term-evolution] or UMB [ultramobile-broadband] connection among multiple LAN clients. Municipal Wi-Fi is another Internet-access candidate in some communities. And then there's WiMax [worldwide interoperability for microwave access], which companies such as Clearwire, Intel, and Sprint are pushing hard. Will the wireless revolution that transformed the LAN have a similar effect on the WAN? And, if so, which technology contenders will rise to the top and why?
While high-bit-rate connections for mobile devices are enticing, they are probably limited by business reasons and not technology reasons. In areas where extensive wired broadband is already deployed, wireless broadband is unlikely to be price-competitive for nonmobile WAN Internet usages in homes and businesses. For mobile devices, we are seeing a hybrid solution for the near future that consists of increased rates for IP [Internet Protocol]-data applications and dedicated transmission systems for video such as DVB-H [digital-video broadcast-handheld]. The new technologists, such as for WiMax, are definitely pressuring the traditional cellular carriers to deploy increased IP-data rates quicker and helping to keep costs down for consumers. But, just like we continually see in the wired world, too many different standards that take too long to solve essentially the same problem become a distraction instead of a catalyst. Consequently, the winner will be the most cost-effective solution to deploy.
Municipal Wi-Fi that is deployed as a mesh has throughput and latency issues for anything but low-bit-rate data applications. That does not begin to address the business model issues where municipalities have struggled to support free services or find the right subscription-cost point. Perhaps that leaves us with something that could be done but shouldn't have been done.
In closing, do you have any other thoughts on networking-related topics we haven't already addressed?
Most of our discussion has focused on the connectivity layer, but we cannot forget the reason we want a large amount of throughput both at home and in mobile situations: the proliferation of digital-multimedia applications and services. We have digital content captured or stored in one device, and we want to see or save it in another device. We have definitely improved from the days of floppy drives to today's high-capacity USB thumb drives for moving content by "sneaker" net. Data content, including e-mail and the Web, has been pretty much solved for mobile and nonmobile devices in the home. But now we need to both easily find and then stream our multimedia content, including both personal and commercial, between all of our electronic devices, such as mobile phones, cameras, computers, network-attached-storage drives, TVs, game consoles, digital-video recorders, MP3 players, and more. The DLNA [Digital Living Networking Alliance] has been steadfastly developing the multimedia-networking infrastructure to enable multimedia connectivity. We are now seeing millions of units shipped that are DLNA-certified. While many may not have heard of DLNA, just like the now-ubiquitous deployment of TCP/IP [Transfer Control Protocol/Internet Protocol] enabled data connectivity, DLNA is also becoming the ubiquitous language for multimedia devices to connect with each other. This [step] will lead to accessing and sharing all-important data and content wherever and whenever we want.
