Altair targets ultra-low power VoWiMAX niche with its first baseband

The WiMAX industry has been enchanted by the open mobile broadband vision of Sprint Xohm and other pioneers, and despite the commercial teething troubles, this remains the key focus of 802.16e marketing and development. Such a model will rely on sophisticated multimedia devices, such as those emerging from Samsung and others, and on mobile internet products like Nokia's Internet Tablet.

The focus on these expensive items has diverted the market's attention from many of WiMAX' original targets, including the provision of an alternative mobile internet network for regions or population groups where 3G+ would be non-commercial, unaffordable or unavailable. But the major growth in uptake of Mobile WiMAX, especially in the next few years, will come from markets far closer to the original remit of proprietary broadband wireless technologies, geared to underserved access and a simpler, lower cost alternative to cellular networks and models. So, according to recent research by Rethink Research, the Indian subcontinent will be the single biggest driver of WiMAX growth in 2008-2010, with some larger sub-Saharan African countries taking this role from 2012. The major services behind the projected multibillion dollar investments by large operators in these regions will not be 4G-style multimedia web services, but simple internet access and VoIP.

These economies and applications will need very low cost devices to be viable for carriers and end users. Initially these may often be shared among families or communities, and will bear closer resemblance to the cheap cordless DECT phones so common in countries like India, than to Samsung's smartphones or even Motorola's ultra-low cost handsets. There is plenty of ongoing work on such devices, notably the venture between Alcatel-Lucent and the research wing of India's Department of Telecommunications. But real products are scarce so far. This is partly because many of the device chipmakers also seem dazzled by the lure of the fully fledged, relatively high margin handset - where they are not fighting for crumbs in the Intel-driven WiMAX laptop arena.

One semiconductor start-up looking to jump into the gap is Altair of Israel, a fabless chip company that has majored on power efficiency and small footprint to target ultra-low cost devices with long battery life (since in many communities there may only be a few power points for recharging). Co-founder and VP of marketing and business development Eran Eshed claims the company's first Mobile WiMAX baseband processor, the ALT2150, consumes only one-third of the power of competitive products. According to him it "represents a true disruption in terms of power consumption, size and cost," and brings WiMAX, for the first time, in line with advanced 3G products in terms of power consumption.

Altair has developed a highly programmable proprietary OFDMA processor (codenamed O2P), rather than integrating a conventional DSP to achieve differentiation. This lies at the heart of the ALT2150, which complies with Mobile WiMAX wave two specifications, along with patent pending ELP (Extreme Low Power) technology. Other features include a high performance MIMO decoder that supports up to 64QAM Matrix-B decoding, and a built-in, patent pending approach to coexistence of WiMAX with Bluetooth and Wi-Fi. The chip is tiny, at 7mm x 7mm.

There is a lot of clever silicon technology in the WiMAX device market though, and there has yet to be the inevitable shake-out of the large number of start-ups fighting for position. A clear target market with short term proven demand, and measurable product differentiation (or at least a technological head start) is the recipe for survival - not just technical innovation. For Altair, its hopes rest on emerging economies, especially in Africa.

According to Eshed, in the real world this technology could lead to a complete VoIP solution at about 35mW in power consumption, with complete costs under $65. The often ignored market on which he has set his sights, as Altair tapes out its first chip, is voice over WiMAX in Africa. "This is a real trend and it's not just about CPEs, because there is often no power supply in the home and so battery devices are needed. We are looking not at $300 smartphones but very basic WiMAX-only cordless handsets for the whole family, charged at a kiosk," he explained. In such a product there would be a separate Mips-type applications processor for VoIP, to reduce cost and increase efficiency. In a closed system, such as what most operators will be happy with in the early stages in developing economies, Eshed says Altair techniques can also improve cell edge coverage and range.

Eshed believes that traditional handset makers will not be making these devices. Instead, he is looking to Taiwanese and Chinese ODMs with expertise in voice over Wi-Fi. The Altair architecture can also be applied to voice over WiMAX CPEs, which Eshed sees as being most relevant in India and Latin America, which need low cost products but generally have more availability of power supplies. He claims a residential VoIP CPE based on the ALT1250 has a similar cost to a data-only product based on chips from a rival (France's Sequans is the competitor Eshed most often cites in his comparisons, but also is the one he seems to respect most among the start-ups). By contrast, he sees most of the small device-side chip players distracted by the laptop card "bloodbath."

Like most WiMAX silicon players Altair is also looking towards LTE, and claims its proprietary OFDMA processor was specifically designed to be easily adaptable to any OFDM-based system. The creation of this core and its surrounding tools as an alternative to the safer, quicker route of integrating DSPs, represents a huge investment and risk for a start-up. However, Eshed claims it not only gives the ALT1250 and its successors their edge, but also makes the architecture flexible to support future changes to the 802.16e specifications, or other OFDM technologies. The company has been working on LTE for a year, and aims to produce one of the first commercial grade LTE ASICs in the device sector by making "minor" changes to the 1250. Altair is also supplying chips to a customer using an OFDMA network that is neither WiMAX nor LTE. However, Eshed knows that in LTE, Altair will need to partner with a larger baseband chipmaker rather than attempting to go it alone in a market of large customers and large vendors. "LTE will get sucked into the baseband far more quickly than WiMAX," he commented. "Qualcomm, Infineon, Marvell, NXP and so on all have internal developments, so there are not many viable tier one baseband partners, and most early stage LTE companies will fail. WiMAX players will already have strong OFDM field experience and so will be well positioned to step in when some of the tier ones see their internal developments fail, and look to partner or acquire."

While Altair waits for that scenario to play out, it claims not to be in a rush for LTE, having remained financially quite conservative - it has raised just $26m to date and is not looking for more for 12-18 months, and "for us WiMAX will pay the bills nicely for a few years", Eshed insists.