Dale’s Web Log

While the Federal Communications Commission moves ahead with planning for the upcoming 700MHz spectrum auction, the White Space Coalition has submitted a second prototype white space wireless broadband device to the FCC for testing. White space devices could use the so-called white space in the current analog television spectrum (2MHz to 698MHz) to deliver wireless broadband service. Former FCC chief engineer Edmond Thomas (and current technology policy advisor for the law firm of Harris, Wiltshire & Grannis, which is representing the Coalition) told Ars that he believes white space broadband could deliver download speeds of up to 80Mbps, which would make it extremely competitive with fiber-to-the-premises solutions like Verizon’s FiOS networks.

The newest white space prototype is manufactured by Philips Electronics of North America and consists of a TV tuner, a digital processing board, and a PC which provides the UI, control, and signal processing. It’s proof-of-concept hardware intended to demonstrate that it’s possible to sense the presence of TV signals and transmit wireless IP data in a way that does not interfere with TV. According to an FCC filing seen by Ars Technica, the new prototype is capable of picking up analog and digital television signals as well as wireless microphone signals (which operate in the same part of the spectrum). It works similarly to the Microsoft-manufactured spectrum sensing device submitted earlier this year. Microsoft also submitted a transmission device to the FCC for testing which will be used to show that white space broadband transmissions won’t interfere with TV signals.

There are a few screenshots in the FCC submission. It’s quite simple: the user selects a type of signal to scan for, and the application shows the results. If the sensing module picks up television transmission on a particular channel, then that part of the spectrum will not be used for white spaces broadband in that particular area.

The goal of the White Space Coalition is simple: take advantage of unused television spectrum to provide wireless broadband. Although analog television transmissions will cease in February 2009, digital TV signals will continue to use the spectrum between 54MHz and 698MHz. That is a highly desirable chunk of spectrum because the signals can easily pass through walls and other solid objects, giving them a much greater reach than WiFi or even WiMAX, both of which operate in higher frequency bands.

Television broadcasters have vigorously opposed the usage of the white spaces, citing fears that wireless broadband will interfere with TV signals. The current round of FCC testing is designed to ensure that the prototype white space broadband devices don’t cause any interference problems at all. “Like the personal/portable prototype devices previously submitted by Microsoft on the Coalition’s behalf, the Philips prototype is designed to demonstrate that operating parameters set forth by the Coalition… will provide incumbent licensees in the television bands with the interference protection to which they are entitled,” reads the FCC filing.

The White Space Coalition is comprised of Dell, EarthLink, Google, HP, Intel, Microsoft, and Philips Electronics. The FCC should conclude its testing of the white space broadband prototypes in July and the first rules governing the use of the spectrum by wireless broadband devices should be released in October 2007. Once that happens, the IEEE will likely begin the work of standardizing the tech. If all goes as planned, white space broadband service could begin in the US as soon as February 2009.

Cisco Systems Inc. is now looking at the prospects of several of its networking divisions introducing WiMax products, according to industry sources, reversing a sometimes combative stance on the emerging wireless technology from the company.

“The wireless, cable, and Linksys groups are all looking at WiMax,” a source tells Unstrung. “These have different motivations and different products.”

Cisco will face the eternal conundrum that it always faces when entering a new wireless market, the source adds: “The question will be, do these internal groups do the work… or who will they buy?”

Another source, however, says at least some of the development is internal and the company is already working on WiMax. This could result in additional WiMax capabilities for its municipal networking offerings.

In the past, WiMAX Forum member Cisco has pooh-poohed the technology’s chances of success as a wide-area wireless access technology. Most notably, when CTO Charlie Giancarlo said the business case for WiMax was “not compelling” in November 2004. The firm has softened its stance a little since then but still has a white paper on its site explaining why it won’t build WiMax base stations.

The company’s official stance on WiMax is still fairly muted. “Cisco always looks at different wireless technologies,” allows Ben Gibson, director of mobility solutions marketing at the firm, but he adds: “WiMax is certainly not nearly as far along in the market as wireless LAN.”

Cisco, however, would by no means be the only major networking company to change its position on WiMax as the market evolves. Just recently, Qualcomm Inc. bought into mobile WiMax, while Ericsson AB decided to get out of the market and concentrate on cellular 4G updates.

Certainly there is more support for WiMax from mobile operators and other service providers now than there was in 2004 or 2005 — when Cisco first got sniffy on WiMax. In the U.S. alone, Clearwire LLC and Sprint Nextel Corp. are working on multi-billion dollar WiMax rollouts through 2008 and beyond.

— Dan Jones, Site Editor, Unstrung

WiMAX (802.16) Service Specific Convergence Sub-layer (CS)

The Service Specific Convergence Sub-layer is at the top of the 802.16 protocol stack. It is responsible for performing the following functions:

• Accepting higher layer PDU’s from the higher layer service
• Classifying and processing the higher layer PDU’s
• Delivering the PDU’s received from the higher layer to the appropriate MAC SAP
• Receiving CS PDU’s from the peer and delivering them to the higher layer

At the present time, there are only two convergence sub-layer definitions for 802.16; the ATM CS and the Packet CS. Thus, only these types of traffic can be transported over the 802.16 network. The standard does leave open the possibility of specifying additional CS’s in the future.

WiMAX (802.16) Packet Convergence Sub-layer (CS)

The first Convergence Sub-layer (CS) I will discuss is the Packet CS. While the packet CS is generic in that it supports any number of packet protocols including IP, PPP, and Ethernet, the most common implementation will likely be for IP networks. In addition to the generic functions above, the packet CS also has the “optional” responsibility for suppressing and rebuilding the packet header information to save bandwidth over the wireless link.

So, how exactly is a protocol data unit (PDU) mapped into the wireless network? Well, essentially the entire packet contents are wrapped into a MAC “Service Delivery Unit” (SDU) – i.e. the packet is encapsulated with the MAC SDU header information (see figure below). Essentially, this is an additional field added to the packet identifying the Packet Header Suppression Identifier (PHSI) if used.

Once the packet has been encapsulated into an SDU, the SDU must be classified and associated with a connection identifier (CID) for transmission to the appropriate peer node. Each CID is known as a “service flow”. The process of classification involves matching each packet against a set of protocol specific matching criteria (e.g. ip address, priority, QoS, etc.) to determine the appropriate CID, or service flow, for that packet. If a packet fails to match any of the classifiers defined for the system, that packet is discarded.

Payload Header Suppression

As we all know, there is a lot of repetitive data in the packet headers on an IP network (Source destination addresses, port numbers, version numbers, etc.). There have even been a fair number of efforts to implement compression schemes based on that repetitiveness (Van Jacobsen compression being the most popular). The folks on the 802.16 committee decided it would be a good idea to build a header compression/suppression scheme into the specification. This way, they would ensure that all WiMAX devices that supported Header Suppression would be interoperable.

The 802.16 Header Suppression technique works by having the MAC SDU of the sending node suppress the header information while the receiving node restores it. Simple enough right? Well not so fast, the sending and receiving nodes need to have a method of sharing connection information such that the suppressed header information is known/recoverable by both nodes. That’s where the Payload header Suppression Index (PHSI) and the Payload Header Suppression Field (PHSF) come into play.

As I described above, the transmitting node uses a classification process to assign packets into specific service flows. For systems that support Payload Header Suppression (PHS), the classifiers will include a PHS rule. On the receiving side, the node uses the CID and PHSI to restore the suppressed header.

Included within the PHS function are two options known as Payload Header Suppression Valid (PHSV) and Payload Header Suppression Mask (PHSM). When the PHSV option is enabled, the transmitting node must verify the payload header before suppressing it. With PHSM, the nodes negotiate which header bytes to suppress and which to transmit. For example, the static header fields (IP address, version, etc.) can easily be suppressed, while the more dynamic fields (ACK number, length, etc.) can continue to be transmitted.

Payload Header Suppression Protocol

• A packet to be transmitted over the 802.16 network is received on the wired interface of the transmitting node and forwarded to the Convergence Sublayer (CS).
• The CS classifies the packet and provides as an output an uplink service flow, a CID, and a PHS rule. The PHS rule identifies the PHSF, PHSI, PHSM, PHSS, and PHSV to be used for this packet.
• If the PHSV value is set, or is not provided as a classification output, the subscriber terminal will compare the bytes in the packet header with the bytes in the PHSF that are slated to be suppressed. If they match, the subscriber will suppress the bytes in the PHSF with the exception of those bytes masked by the PHSM. If they do not match, the transmitting node sets the PHSI to 0 and the header is not suppressed.
• The transmitting node appends the PHSI to the packet and transmits it to the receiving node.
• At the receiving node, the packet received over the air interface is forwarded to the service flow associated with the received CID.
• The convergence sublayer on the receiving node uses the CID and the PHSI to recover the PHSF, PHSM, and PHSS.
• The packet is then reconstructed with the original packet header fields and forwarded normally.
• If PHSV was enabled on the transmitter, then the PHSF bytes recovered by the receiver are guaranteed to match what was transmitted.
• If PHSV was not enabled on the transmitter, then there is no guarantee that the PHSF bytes recovered by the receiver will match what was transmitted.

The figure below is an excerpt from the 802.16 specification and depicts the decision logic involved in PHS.



I hope this is useful. I know it’s helping me learn the protocol. Next I will start into the specific packet CS’s covered by the specification (Ethernet, PPP, and IP).

This is the first in a multi-part series of posts I am working on. I am attempting to familiarize myself with the 802.16 protocol standards and the WiMAX forum efforts. I find that the best way to learn stuff is to write it down. I hope this helps others out there, and by all means, if you see and error, please comment and I will correct it.

Introduction

The IEEE 802.16 working group is the standards body developing the specifications for the next generation of broadband wireless networks – both fixed and mobile. The WiMAX (Worldwide Interoperability for Microwave Access) forum is an interoperability initiative that will certify that broadband wireless radios manufactured by various vendors comply with the IEEE 802.16 specification and – more importantly – are interoperable from vendor to vendor through testing.

WiMAX Range and Throughput

The throughput and range of a WiMAX link is very dependent upon a number of factors including; transmit power, antenna gain, directionality, modulation scheme, forward error correction codes, the terrain, density, height of tree cover, presence of hills and valleys, bodies of water, etc. (nothing new here, these are all the typical issues that need to be dealt with in wireless communications). Despite this, it is not uncommon to see statements in the media describing WiMAX multipoint coverage as being capable of extending coverage up to 30 miles from a base station or tower location. In certain (very specific) cases this is true. However, typical operating ranges fall into the 8-10 mile range for line of site (LOS) installations and 4 – 5 miles for non or near-line of sight (NLOS) installations.

Realized throughput is dependent upon many of the same factors that address the achievable link distance. WiMAX supports a number of different modulation schemes and coding rates. Also, it is important to remember though that WiMAX is a shared network service, meaning that this bandwidth will be shared amongst all of the users within a given WiMAX cell – much like cable modem users share their Internet bandwidth with other subscribers in the neighborhood.

802.16 Reference Model

802.16 is essentially a wireless implementation of Layers 1 and 2 of the OSI model (Physical Layer and Link Layer). The Physical layer is referred to as the PHY. The link Layer is referred to as the MAC (Media Access Control) layer. The MAC is comprised of 3 sub-layers; the Service Specific Convergence Sub-layer (CS), the Common Part Sub-layer (CPS), and the Security Sub-layer (SS).

The Service Specific Convergence Sub-layer provides the interface between the WiMAX network and the higher level protocols. Data received from the higher layer protocols is mapped into the appropriate WiMAX identifiers for transmission over the network (for example, Ethernet Addresses are converted to WiMAX station identifiers). Currently there are only two CS interfaces defined; one for ATM traffic, and one for packet (IP) traffic.

The MAC CPS performs the bulk of the MAC layer processing. It is responsible for managing system access, bandwidth allocation, and connection management.

The MAC Security Sub-layer provides authentication, secure key exchange, and payload encryption.

The WiMAX Physical Layer (PHY) actually consists of multiple sub-specifications, each one dependent upon the application and frequency spectrum to be utilized. There are three basic frequency bands to be concerned with:

• 10 GHz to 66 GHz Licensed Frequencies
• Licensed Frequencies below 11 GHz
• Unlicensed frequencies below 11 GHz (primarily 5 to 6 GHz)

Coming next….a look at the Convergence Sublayer of the MAC.

An excellent article about the history of the 802.11 standards and lessons learned that should be applied to the future. The original article can be found here.

To better understand the politics of 802.11n, it is necessary to first provide some perspective on how the industry got to this point.

The first wireless network standards were approved in late 1999 by the IEEE as part of the 802.11b effort. Those standards featured two major technologies to distribute packets over the radio spectrum using spread spectrum methods that are still in use by most wireless networks.

Shortly after, the 802.11a standard was ratified, which used orthogonal frequency division multiplexing (OFDM) methods to enable higher data rates. Instead of using the same 2.4 GHz frequency band as existing 802.11b products, however, the 802.11a standard operates at in the 5 GHz frequency range. This allows wireless designers to take advantage of a greater number of non-overlapping channels for transmitting data.

With the benefit of hindsight, we can see now that 802.11b’s popularity created a powerful legacy ecosystem and established the need for backward compatibility as new wireless protocols emerge.

Lesson 1: Support legacy 802.11b.
The WLAN industry has since learned the importance of supporting legacy 802.11b devices. While it seemed like a good idea at the time to establish a new and potentially higher-performing frequency band at 5 GHz, this incompatibility was ultimately a disadvantage for these products.

While both 802.11a and 802.11b were important efforts, neither could handle the demands of multimedia applications, such as streaming audio and video. Therefore, another effort began to extend these protocols to support higher throughput and lower latencies. That effort turned into the 802.11g standard that was ratified in 2003, which applied the frequency division techniques of 802.11a but used the original 802.11b radio frequencies (see Table 1 below).

Despite lackluster adoption, 802.11a product lines weren’t developed for naught. Working on 802.11a taught engineers how to build radios that operate on different frequency bands.

Many multi-frequency products now combine a/b or a/g radios together, and support clients that operate on either frequency. 802.11n products will have multiple-frequency support built-in, so clients can transmit and receive on both the 5 and 2 GHz spectrums simultaneously, boosting bandwidth and throughput and taking advantage of the larger number of channels and more efficient radio transmissions at the higher frequency range.

Lesson 2: Channels matter.
The number of separate transmission channels in the 802.11b/g frequency range is effectively three: channels 1, 6, and 11 are the only ones that don’t overlap with the others.

In radio-rich environments, such as a major downtown urban core, there will be plenty of interference from neighboring wireless networks. To help improve throughput, the 802.11n task group is not only using the 5 GHz spectrum to increase the number of channels, but is also considering doubling the size of the channel itself. Thanks to improvements in channel utilization, most 802.11n products are expected to deliver 100-300 Mbps data rates.

With 802.11a and 802.11b, it took several years between the standard ratification and products appearing on retail shelves. The 802.11g effort was the first time that products were introduced in advance of a final standard.

The success of that strategy has influenced how companies have come to market with the 802.11n chipsets in the past year, and one of the reasons there are so many “draft-n” products sold by the major manufacturers.

By developing products in tandem with the evolving specification, many vendors hope to shrink time-to-market for 802.11n products to less than a year. This brings up the next lesson:

Lesson 3: IEEE and Wi-Fi Alliance need to work in parallel.
In the past, engineers in the IEEE working groups developed a draft standard. Once that was finalized, engineers in the Wi-Fi Alliance would start putting together a test and certification plan to ensure interoperability between various implementations of the standard. But as vendors move more quickly and the market becomes more complex, these two groups need to work in parallel.

That is now happening with the 802.11n standards process. The Wi-Fi Alliance agreed last November to start developing 802.11n certification processes in parallel with the IEEE 802.11n working group.

This also helps because the Alliance needs more time to test the implementations submitted from each vendor due to the complexity of the 802.11n standard. Greater inter-organization collaboration may remove issues and differences, such as those that arose with 802.11g: the IEEE made anything over 24 Mbps optional, but Wi-Fi certification required products to support 54 Mbps rates. Speaking of these higher rates, this brings us to the next lesson learned.
Lesson 4: Latency matters.
All of these audio/video applications are placing more importance on latency rather than throughput, which is another factor driving the draft 802.11n standards efforts that combine improvements on both fronts.

While greater bandwidth is critical, advanced wireless applications will not work acceptably on today’s networks without reducing latency times. The difference is important: latency is the total round-trip time that information takes—going from client request to server response and back to the client.

Even the fastest networks suffer from long latencies, which can wreak havoc with audio synchronization, or give users the feeling that “nothing is happening” when they hit the Enter key on their PC browsers. With 802.11n, there is an opportunity to make huge improvements in both latency and bandwidth.

There are several improvements that will help reduce latency. Latency matters most in synchronizing audio applications, particularly with streaming video and with VoIP situations. Delays of a just a few milliseconds can add up over various network router links and result in a mismatched picture to the sound, or turn a phone call with crisp quality into a jumbled mess.

One of the more important is frame aggregation, which groups data packets into larger frames to minimize packet overheard. While throughput is still important, without low latency times, many of these newer applications would not work acceptably on today’s networks.

Speaking of the higher data rates supported by 802.11g, here is another lesson. The high end of 802.11g networks sounds very promising, supposedly delivering 54 Mbps data rates. However, what is really going on at these speeds is that there is an almost a 50% overhead that is created to sustain the highest throughput.

Lesson 5: Minimize packet overhead.
Going from 24 to 54 Mbps doesn’t buy twice the performance that the raw data rate numbers imply. Any new protocol must do a better job at managing packet overhead than 802.11g, which uses nearly half of its available bandwidth for overhead at the highest data rates.

During 802.11n development, engineers worked hard to reduce overhead and eliminate turnarounds wherever possible. One of the most prolific improvements is frame aggregation.

Instead of sending a single data frame with its overhead, the transmitting client combines and sends a series of frames with a single overhead frame without waiting for each packet to be individually acknowledged.

As a result, an 802.11n client will send more data in a given time period ” which makes the transmission more efficient.

Figure 1: How frame aggregation improves packet efficiencies

Lesson 6: Bring consensus whenever possible.
As the wireless technology impacts a greater number of manufacturers and vendors, relevant working groups get bigger and it becomes more difficult to gain consensus on technical aspects of a standard. By offering an alternate venue for drafting the 802.11n specification, the Enhanced Wireless Consortium (EWC) enabled vendors to put aside their differences and move more quickly towards an agreed-upon standard.

As previously mentioned, the 802.11n specification is much more technically complex than earlier 802.11 standards. One of the byproducts of this complexity is a longer input and approval cycle for new wireless standards. As the membership of various 802.11n working groups had grown, it became harder to achieve consensus.

Late last year, Broadcom and several competitors united stakeholders from competing standards camps to iron-out technical issues and achieve consensus more quickly. This new body, called the Enhanced Wireless Consortium, brought together talented engineers from the major Wi-Fi chip vendors, device manufacturers and PC OEMs to put differences aside and develop a draft specification.

By jump-starting consensus, the EWC laid the groundwork for moving the 802.11n standard towards finalization. The EWC draft was universally accepted by the IEEE 802.11n working group earlier this year, and is on track to be approved early in 2007.

Lesson 7: It isn’t only about PCs anymore.
It is clear that PCs aren’t the only wireless devices that matter. Any new wireless standard needs to take into account scores of other consumer electronics that will find their way onto home and corporate networks. The 802.11n standard must specifically embrace these devices and support their operation on formerly all-PC networks.

One of the major obstacles for the digital home is putting in new wires to connect everything. Some estimates indicate that Wi-Fi is used to connect devices in half of all homes with broadband connections. This number will continue to climb as the wireless products become more capable. The increasing popularity of wireless networks will spur an increase in the number of applications that will run over these networks.

Indeed, as wireless networks become more pervasive, all of us will see more Wi-Fi enabled printers, music and video devices, cell phones, digital cameras, etc. This new generation of products will enable the wireless transfer of images music and video content that are stored on media servers and media players like Apple’s iPod, and Voice over IP telephones.

All of these devices want to share content, applications, and IP addresses wirelessly on the same network. The shear volume of wireless-enabled devices will also place more demand on wireless throughput, requiring higher performance networking protocols. The earlier 802.11 standards were designed exclusively for PC clients, while the draft 802.11n standard has extended wireless support to these new devices. As more consumers build libraries of digital music, movies and photos, they want to access their digital content on a wide range of devices in various rooms around their homes.

Having a higher-performing network will enable these sorts of options. But this also presents challenges for electronic designers who must now package wireless solutions in different form factors to support these non-PC devices, and must take into account different battery and power consumption profiles that these devices make use of.

Another application for higher-speed wireless networks has to do with network-attached storage. As prices drop and storage needs increase, these units are being pressed into service for video servers and digital music. Network attached storage devices will require reliable, high-bandwidth connections to stream pre-recorded high-definition TV shows, music videos and full-length feature films to televisions and computers throughout the house.

One final lesson is to incorporate the multiple antenna designs that have been developed for enterprise-class access points into the consumer-grade wireless networks. Many of the 802.11g products came to market with multiple antenna designs, and of course each vendor uses a different set of electronics and designs for how these multiple arrays are packaged.

The idea is that the more antennas, the longer the range that your radios can send and receive signals. While that is true in an ideal situation where a single access point is talking to multiple clients scattered around the landscape, the real world is a much harsher RF environment with multiple access points and interference from lots of radio emitters like microwave ovens, cordless phones, and older Wi-Fi networks.

Lesson 8: Multiple antennas matter.
In the quest for higher data rates, having more than one antenna and more than one radio stream matters for delivering better throughput. The 802.11n draft is the first IEEE standard to support these innovations.

Despite all of these history lessons, the industry still hasn’t finalized the 802.11n standard. It will be at least the beginning of 2007 before it is ratified.

Why so long? There are several issues. First, the specification has an unprecedented collection of new technologies, with more variety and depth than in any previous 802.11 standard.

Second, the wireless market has expanded greatly over the past several years: it is hard to find a laptop sold today without built-in Wi-Fi features, and users have come to expect wireless access at their hotels, airports and in major downtown business districts.

The increased market means more engineers present at the various working group meetings and the longer it takes to reach consensus. This was one of the reasons Broadcom and others took a separate group of experts outside of the IEEE process to chart a common ground for the 802.11n standard.

Third, the diversity of new wireless products beyond just PCs means that new test plans and processes have to be put into place to certify them for the new standard.

Finally, there is the issue of supporting earlier 802.11 standards and making sure that new products play well with older ones. This increases the testing and certification process as well. One of the things the industry is addressing in the 802.11n process is to ensure that new innovation doesn’t harm existing 802.11 networks.

The introduction in 2003 of products based on 802.11g specifications provided large numbers of consumers with the benefits of wireless networking for the first time. With its ability to enable sharing of Internet connections and printers without installation of new cables throughout the home, Wi-Fi has emerged as the fastest growing technology for home data networks. And with the coming of 802.11n, more capable networks are enabled with higher performance and richer support of multimedia applications.

Table 1- Summary of Different Wireless Networking Approaches

Notes: (1) OFDM - Orthogonal Frequency Division Multiplexing, “+” indicates improvements that are part of 11n draft standards with wider bandwidth and higher code rates; DSSS = Direct-Sequence Spread Spectrum; CCK = Complementary Code Keying.

About the author
David Strom is an author, consultant, podcaster, and speaker on a wide variety of networking and communications topics. He was the founding editor-in-chief for Network Computing and created the DesignLine series of sites for CMP Media. He can be reached at david@strom.com and is based in St Louis.