WiMax (the World Wide Interoperability for Microwave Access) technology was developed by the IEEE 802.16 Working Group as a standard for Metropolitan Area Broadband Wireless (MABW). So far so acronym heavy. But, with conformance and interoperability testing for the 2004 WiMax Fixed version now in place, and silicon expected during 2006 for the soon-to-be-published 2005 WiMax Mobile version, a potentially major new broadband access technology is now touting for business.

This could be a big headache for many operators and service providers, already confronted with a range of broadband technologies -- fiber, the innumerable flavors of DSL, and 3G mobile, for example. WiMax has many attractive features, especially to new-entrant competitors, because its wireless basis makes it relatively infrastructure light and flexible compared to wired technologies, and it could hit the bulls-eye of fixed/mobile convergence, but it is new and largely untried.

So it’s important to see how WiMax technology can be deployed to support broadband access services that matter to telecom operators, now and in the future. This means looking at the kinds of application it can support, the availability of spectrum, and likely product timelines. Equally important is product performance, and what can be done to improve it, and to lower costs, which are crucial to the WiMax business case.

Topics covered in this report are:

Webinar

This report was previewed in a Webinar sponsored by Intel Corp. (Nasdaq: INTC){dirlink 2|39} and Proxim Corp. (Nasdaq: PROX). It may be viewed free of charge in our Webinar archives by clicking here.

— Gabriel Brown, Chief Analyst, Light Reading Insider

For an exhaustive look at how WiMax will fit into networking plans for 2005 and beyond, check outWiMax: From Development to Deployment, to be held at the Westin Times Square in New York Cityon November 16, 2005.

Presented by Light Reading, in association with Unstrung and Heavy Reading,WiMax: From Development to Deployment will be the first conference and exhibition to addresspressing WiMax questions and provide a clear picture of how the market is evolving.

For more information, click here.

Sponsorship opportunities are still available. Direct all inquiries to: [email protected].

WiMax technology can be deployed to do several different things, as shown in Figure 1. They are:

Fixed broadband is a fixed usage model, such as backhauling WiFi hotspots for enterprises, or copper DSL replacement with wireless DSL. It is generally about small-business or enterprise connectivity, and there is a fixed (usually building-mounted) endpoint. This means that the operator’s network planning is fairly straightforward, because it knows where the clients are going to be for the purposes of capacity and base-station planning.

Fixed-to-portable broadband is a nomadic type of usage, where clients are not being used while in motion, but can nevertheless move from place to place within a service area or zone. Typical examples are enterprise campus usage, or indoor equipment, such as smaller equipment that the user can put in their pocket and take with them from one place to another.

Mobile broadband is a full-mobile usage model, where clients are fully mobile -- moving around from place to place and roaming between networks, and being used while in motion. This is very much about future notebook/handheld-device integration.

The IEEE 802.16 specifications on which WiMax rests are evolving to address these three primary stages of deployment. The first specification, 802.16-2004, was designed to address the fixed usage model. It is followed by 802.16-2005 (or 16e), which addresses the fixed-to-portable and mobile usage models.

The bottom of Figure 1 shows how different types of operators could use WiMax. The blue arrow shows fixed operators that may be expanding or moving to more mobile-type applications over time, and the purple arrow shows mobile operators that may be moving more towards traditional DSL models. So a WiMax environment would allow operators to move in both directions, as well as new entrants and ISPs competing as well. Overall, WiMax could be a significant enabler of telecom competition. Wireless ISPs (WISPs) are expected to be some of the earlier adopters for the WiMAX Forum certified systems, as WiMax brings more choice and easier integration for those who want to enter the WISP market.

Application ClassesA crucial question for operators is what kinds of broadband applications WiMax will support? The answer is many. Because WiMax offers QOS, a single system can support multiple different types simultaneously. This gives operators deployment flexibility.

The WiMAX Forum Applications Working Group (AWG) has specified five initial application classes, as shown in Table 1, and initial WiMAX Forum Certified systems will be capable of supporting these five classes simultaneously.

Table 1: WiMax Application Classes

The different classes all have different requirements for three key network parameters: bandwidth, latency, and jitter. This complicates deployment.

Bandwidth looks at how much data capacity an application is consuming. Applications like interactive gaming and VOIP use very little bandwidth, but applications like downloading or streaming multimedia can be very bandwidth heavy.Latency is a measure of the time it takes for data to make a round trip of the network, and is central for applications that use real-time communications, like voice and video. The WiMAX Forum has established guidelines for the maximum amount of latency or delay that would permit these applications to operate acceptably. Actual system latencies, with WiMax, could be as low as single digits of milliseconds.

Jitter is the variability of the information received as it relates to time. This is key for such things as synchronizing audio and video for an acceptable user experience. The WiMAX Forum has also established limits to jitter to ensure the delivery of high-quality services.

WiMax technology was developed by the IEEE 802.16 Working Group as a standard for Metropolitan Area Broadband Wireless, initially at higher frequencies. After several revisions, the 802.16-2004 specification (also ETSI HiperMAN) was finalized in 2004, and is now commonly known as WiMax Fixed.

This standard specifies a radio interface that uses adaptive modulation and orthogonal frequency-division multiplexing (OFDM) to help reduce the impact of multipath interference. This makes WiMax suitable for non-line-of-sight environments, such as urban areas or rural areas with lots of trees.

802.16-2004 has frequencies specified as sub-11GHz, scalable channel widths of 1.75 to 20MHz, 256-carrier OFDM , and frequency-division duplexing (FDD) or time-division duplexing (TDD).FDD essentially means that there are two full radios at each end of the connection. One of the radios is designed to transmit on one frequency while the other is receiving on a totally separate frequency. TDD uses time to determine when information is being sent or received over a single channel. A feature of TDD is that it can be weighted so that, say, 75% of the time will be used to transmit information down to the users, and the other 25% to receive it from them, thereby adapting network capacity to usage patterns.

Another important feature is that the 802.16 MAC offers deterministic QOS. This is crucial, because it makes it practical to offer services such as voice and T1 line.

In December 2005, the IEEE expects to publish a new WiMax standard dubbed WiMax Mobile, 802.16e/-2005. Although this has some commonality with the 2004 standard, it is not backwards compatible with it.802.16e/-2005 has frequencies specified as sub-6GHz, and scalable OFDMA with 128, 512, 1024, or 2048 carriers.

The point of the differences between the two versions is fiercely debated within the WiMax and wider wireless communities. The nice and cosy point of view is that each version is designed to accommodate the very different needs of fixed and mobile operation. The more controversial point of view, and arguably more realistic, is that scalable OFDMA (802.16e) will emerge as the de facto WiMax standard for both fixed/portable and portable/mobile deployments. How this debate will play out is yet to be determined. See: All Hail OFDMA!

Spectrum Options802.16 is a very broad specification, running through many frequency bands. There is no single frequency or channel width or multiplexing scheme specified, which is both a strength and a weakness when it comes to creating a global mass market for WiMax products.

Table 2 covers the most likely options for WiMax systems in the licensed and unlicensed spectrum bands. Generally speaking, it is better to have spectrum in the lower bands (for longer ranges and building penetration), although this can be offset by the sheer amount of spectrum potentially available in the higher bands.

Table 2: WiMaxSpectrum Snapshot

Initially, the lead band is expected to be the 3.5GHz frequencies that will be allocated to service providers in many international markets. Typically, these are licensed for fixed use. In the U.S., however, the lead market is likely to be the unlicensed upper 5GHz band, where there is 125MHz of spectrum available. By virtue of being unlicensed, this is likely to be an exciting and fast-moving area.

Over the next two or three years, systems will also be introduced in the 2.5GHz bands. In the U.S., this spectrum has already been allocated, while in much of the rest of the world it has been earmarked for IMT or 3G expansion. There appears to be a consensus that these expansion bands will be licensed on a technology-neutral basis. This opens the door for mobile WiMax in many countries.

The 3.65–3.7GHz band is potentially very interesting for smaller wireless ISPs. This spectrum is about to be released in the U.S. under a licensing-lite regime, whereby it is relatively easy and low cost to get a permit from the regulator to use the spectrum. It looks possible that this will become a de facto WiMax band, although it may involve some changes to the protocol for this to happen because the licensing terms look set to specify contention-based protocols similar to 802.11 (WiFi), whereby radios back off from one another if there is co-channel interference.

Another intriguing area is the lower spectrum bands, mainly the 700MHz band, currently the UHF band for the upper TV channels. Many consider this to be underused, or at least not used heavily. Potentially, much of that frequency spectrum could be opened up in the U.S. and perhaps in other countries. It is not an area that has been assigned yet, but the FCC is looking at it. It would certainly be good a spectrum for WiMax use, as it supports long-distance propagation and will penetrate buildings well.

There is a huge amount of choice in the 802.16 (and Hiperman) specifications, including an immense range of frequency bands (these allocations will change from country to country), half-duplex or full-duplex operation, and different types of modulation. These do not cover the different regulatory requirements in different parts of the world.

For practical equipments and deployments there is an obvious need to narrow the focus of the IEEE technical specifications and to enable broad interoperability from multiple vendors. This has been done successfully in wireless LANs by the Wi-Fi Alliance, and is now being done in wireless broadband access by the WiMAX Forum.

Launched in 2003 by a group of vendors including Intel Corp. (Nasdaq: INTC){dirlink 2|39}, Nokia Corp. (NYSE: NOK), and Proxim Corp. (Nasdaq: PROX), the WiMAX Forum now has over 300 company members, ranging from service providers to silicon vendors. Its aim is to define conformance and interoperability for 802.16 and Hiperman. This is a wider aim than for WiFi, where only interoperability is tested.

The Forum establishes a profile, which defines the different characteristics and baseline functionality, and systems vendors can develop products that match those profiles and submit them for certification and interoperability testing. It needs a minimum of three vendors to test for a specific profile for conformance and interoperability.

The test cases are defined by the Forum and scripted by European Telecommunications Standards Institute (ETSI). An independent lab (Cetecom SA) carries out the testing (vendor product submissions have been accepted from July 2005).

Profiles will continue to evolve as 802.16e is ratified and certification is defined.

Table 3 shows existing and some planned future profiles. All use common capabilities, such as OFDM and point-to-point/multipoint configurations. They vary by frequency band and duplexing method, and channel size. Initial testing is in the 3.5 and 5.8GHz bands.

Table 3: WiMax Profiles: Point-to-Multipoint, 256-Carrier OFDM

The 2.5GHz band has also been approved, but it is expected that systems will probably be developed for this band as mobile applications, and so the WiMAX Forum has not yet completed the profile for 2.5GHz.

To speed the introduction of WiMax certified products, about 75 test cases will be exercised on initial products. However, about six months after the first certification wave of products, the test cases will be increased to include areas like security. After about an additional six months the Forum expects there to be over 200 mandatory test cases that will also encompass QOS.

This process of expanding the testing scope is similar to that in wireless standards, such as GSM and WiFi. It is expected that over the first year there will be more test cases for WiMax systems than there are for GSM in 2005.

Product Timelines

Intel announced silicon for 802.16-2004 products early in 2005, and these are incorporated in systems undergoing the first round of WiMax certification. WiMax-certified systems will appear in late 2005, and network deployment will start then. Note, however, that many pre-WiMax 802.16 systems have already been deployed around the world, and use the same technology as officially certified WiMax equipment.

802.16e/-2005 silicon will probably not appear until mid-2006 or in the later half of that year, with WiMax certification and system availability following. Network deployment could start in late 2006 or during 2007. Note that prototype PCCards and handheld devices incorporating OFDMA user-terminal ASICs and base-station equipment incorporating OFDMA baseband processing are already available for vendor and operator testing. The first networks based on such technology could be launched before the end of 2005 in South Korea.

A commonly held misconception is that WiMax can deliver massive broadband performance at very low cost. One way of summarizing this view would be “WiMax will deliver 70 Mbit/s over 70 miles, at 70MPH, with a $70 CPE device.” This is nonsense.

There are always trade-offs. Long ranges and high bit rates require higher radio powers, better antennas, and bigger and better batteries -- all of which cost money.

As a real-world example, the French operator Altitude Telecom{dirlink 5|7}, has reported measured throughputs for its 3.5GHz pre-WiMAX (so uncertified) 802.16 radios for symmetrical TCP/IP transmission. On line-of-sight paths it achieved 10 Mbit/s over 10km, but this fell to 4 Mbit/s over 5km for non-line-of-sight paths (and the degree of obstruction was mild, being limited to occasional trees or part of an occasional building). The network operates in FDD mode in 3.5MHz channels with outdoor CPE.

Figure 2 and Figure 3 show the results of modeling by the WiMAX Forum that tries to address some of these issues. The system is assumed to be operating at 3.5GHz, with 3.5MHz channels, FDD, over near/non-line-of-sight, and with outdoor CPE.

Figure 2 is for a single RF carrier. From a deployment perspective, the key point is that the capacity shown must be shared by all subscribers within that RF carrier sector, so operators need to think about acceptable contention ratios and so forth, which obviously differ for different classes of service and QOS.

Figure 3 shows the aggregate performance for a 6-sector cell site. Again, this capacity is shared among all the users of the cell. So, at 3km with a suburban cell, the carrier can have 6 × 60° sectors, with each sector having 5-Mbit/s throughput, for a total cell throughput of 30 Mbit/s.

These throughput figures show why the strong QOS capabilities of WiMax are vitally important. Some applications have particular SLA requirements, so a customer will not want a VOIP call to contend with a large file download, for example. That WiMax can offer QOS is a strong advantage over other technologies such as DSL and WiFi.

Another point from a deployment perspective is that WiMax can in some ways be more complicated than a cellular network when planning for coverage an capacity. Although the subscribers do not move around in a WiMax Fixed system, operators cannot necessarily easily move or resectorize the base stations, because this may involve having to revisit all the installed CPE antennas and realigning them. This may not always be necessary, but it is something that operators must bear in mind.

Unsurprisingly, these are just some of the reasons that major carriers (such as AT&T Corp. (NYSE: T){dirlink 5|11} (NYSE: T) are carrying out large-scale WiMax trials ahead of the commercial launch of services.

In comparison to operation at 3.5GHz, a 5.8GHz WiMax system using a 10MHz-wide channel in TDD mode (that is, in half-duplex mode), the base station data signaling rate will be 35 Mbit/s, and will support customers to about 3 miles line of sight. That same base station can fall back to as low as about 7.5-Mbit/s data signaling rate to support customers as far out as about 18 miles line of sight. Near and non-line-of-sight deployments are also supported, but at significantly lower ranges than for line of sight, regardless of the frequency band.

There is likely to be a lot of variation in base-station equipment, tailored to the deployment scenario. Significant factors will be price, form factor, capacity, and performance. Two major types of base station will be:

Initially, form factors will be very similar to those of the proprietary broadband wireless systems of today. Base-station sectors will be perhaps the size of three or four notebook computers stacked end to end, and weigh of the order of 40 to 60 pounds. CPE will certainly be much smaller, and probably have more of a DTV satellite form factor initially.

A key factor in design selection is base-station positioning, the ideal being good, tall broadcast points that are directly in the middle of the user population the operator wants to serve. More often, the operator has to use a mountain top, communications tower, or tall rooftop located somewhere that is noncentral to the target population. So operators will have a balance between big-stick cellular-type deployments that serve all directions, and have a lot of equipment costs associated with them, versus smaller standalone base station sectors.

Experience to date from Proxim suggests that, for fixed clients, full 360° coverage is not often required, even from a prime central location, 180° being the average. With WiMax Mobile this is likely to change, so operators need to have the flexibility to cope with this.

Vendors will naturally introduce enhancements to improve performance, such as smart antennas for non-line-of-site operation, MIMO antenna systems, and so on. There should be a vibrant subsystems market based on open base station architectures.

Smart antenna technology is a particularly interesting development because it can be used to make indoor CPE or indoor self-install modems much more viable, especially towards the edge of the cell, where signal strength is weakest. This is important because the cost of CPE and the cost of installing them are absolutely critical to the WiMax business case, and cutting out truck rolls is worth a lot. Figures from the WiMAX Forum suggest that smart antennas could, in some cases, effectively quadruple the non-line-of-sight and indoor self-install ranges obtainable at 3.5GHz compared to standard antennas -- from 0.5km to 2km for indoor self-install, for example.

It is widely expected, therefore, that multiple antenna schemes will be a key attribute of WiMax systems. Table 4 from last-mile wireless vendor SOMA Networks Inc. summarizes the options.

Table 4: Options for Multiple Antenna Schemes

Figure 4 gives a view from Intel on likely CPE price trends over the next few years. CPE cost is crucial to the success of WiMax, and has to fall, especially to address emerging markets, which are seen as one of the major areas for WiMax. If CPE is too expensive, it will greatly limit the number of people who will use it, limiting essentially to enterprise customers or premium types of consumer.

The basic dynamics of Figure 4 state that in 2005 prices fell as the 802.16 standard was increasingly adopted, hence driving more volume. Also vendors can leverage common design in terms of CPE design. However, there is still the installation/truck roll, which adds some cost.

Certification and interoperability come into play at the end of 2005, and further drive the cost down, because of more competition between vendors and also higher volumes, which drive greater economies of scale.

In 2006 truck rolls are eliminated because of have indoor CPE. In 2007 PCCards arrive, and, in 2008, there is eventually notebook integration of WiMax along the lines of Intel’s Centrino WiFi product. Each of these steps drives the CPE costs down further because of increasing levels of integration. Note that many vendors believe this cost reduction curve is not realistic, and is too aggressive by the order of at least 6-12 months -- in particular the likelihood of $100 self-install indoor modems being available by mid-2006 is considered low.

WiMax Enterprise Services Deployment

There is potentially a large market for WiMax in enterprises. Often businesses are connected by leased lines from the local incumbent, and a common complaint is that such T1/E1 lines are often expensive. This provides an opportunity for wireless service providers to come in with a better offer.

Table 5 is an example from TowerStream Corp.. The offer is a private-line T1 1.5Mbit/s service with SLA at a competitive price, but with a twist. The customer is allowed to burst to a higher rate on a best-effort basis should the necessary bandwidth be available at the time in the customer’s sector. Several other wireless providers are using such extra burstable bandwidth to attract customers that would otherwise use a wired T1 leased line.

Table 5: TowerStream Wireless Broadband Offering

The keys for a case like TowerStream is flexibility and scalability. Such operators are going in to service metropolitan areas that are already heavily wired physically with T1 and fiber optic cables, so they have to find a new way to be competitive. Their competitiveness is driven from the fact that they are wireless, so they can usually offer service within 24 to 48h, and they can scale that service if the customer needs additional bandwidth.

This model could be successful, at least in the short term. Note, however, that the business case is partially founded in the perceived high price of leased lines, rather than the cost of delivering leased-line services. There is obviously a risk that any wireless provider that becomes too successful would see wireline competitors reduce prices and/or increase bandwidth in response.

WiMax Consumer and Enterprise Urban Deployment

The WiMAX Forum has prepared several case studies for business cases for different WiMax deployments. Although they are theoretical and inevitably partisan, the overall assumptions and outcomes do not appear wildly unreasonable. A typical example for a mixed consumer and enterprise urban deployment is a 4-year business case, based on about 3.5% penetration in a 136km² coverage area with 48 base stations 1.8km apart, 12 base stations being added per year.

It shows the operator beginning to pay back capital costs after about four years of operation, and moving into profitability relatively quickly thereafter. By year 5 there are 4800 enterprise and 42 768 residential customers, generating an annual cashflow of $8.7 million for a cumulative capital expenditure of $14.6 million.

Download these case studies here.