LONDON The past few weeks has seen frenetic activity in the still largely untested femtocells business --also known as home base stations or 3G access points.
A group of about 25 companies including chip suppliers, mobile network operators and the developers of these mini base stations have joined together to form the Femto Forum, (of which only seven would go public); several operators have revealed they have put out requests for proposals to equipment vendors; and hardly a day goes by without some of the key players announcing they are partnering to bring femtocells to market.
Earlier this month one of the first international congresses focusing on the technology, organized by Avren Events, and held in London, attracted over 250 delegates.
Still, it is clear there are significant technical, regulatory and commercial hurdles to be cleared before the widespread deployment of femtocells, which can provide enhanced voice and data coverage in the home for, typically, up to 6 users on their existing 3G handsets.
Femtocells represent the first real threat to the increasing dominance of Wi-Fi routers in the home and offer the prospect of an all-IP approach to increased coverage, while backhauling cellular traffic over a broadband connection. As such, they could pave a way for cellular operators to beat off the threat of loss of revenue from voice over Wi-Fi, where calls are handed off to a Wi-Fi network and then carried back to the mobile network over a cable broadband or DSL connection.
It would seem that cellular network operators should be beating a path to the femtocell vendors. But apparently they are not yet convinced of the cost efficiencies femtocells offer and are confused by the many options for integration into the core network -- with the three key versions being lub over IP, UMA (Unlicensed Mobile Access), or based on IP Multimedia Subsystem (IMS) and Session Internet Protocol (SIP) standards-- nor even agreed whether the business case is predicated on anticipated much higher data usage or better voice coverage.
Industry analysts and vendors, not surprisingly, are convinced femtocells are set to make the mark by the middle of next year, and then soar to large volumes quickly.
Vincent Poulbere, principal analyst at Ovum, suggested a slow build-up in 2008 in Western Europe to about 7 million units shipped in 2009, rising to 17 million in 2011.
Meanwhile Stuart Carlaw, research director at ABI research, said: "It's a high risk, high reward environment for now, but we see 102 million users by 2011 on 36 million femtocells. But that depends on operators converting the trials they tell us they are conducting to major deployments.
"If things do not go as smoothly as anticipated, and the buzz and expectation that is being raised does not materialize, things could turn out bad, particularly for the many innovative, start up companies that are currently driving the technology," he said.
Carlaw added the introduction of open standards will be important for success and thus welcomed the Femto Forum's target of focusing on that in the first instance. "The industry really needs to focus on economies of scale and not get fragmented," he said.
From a semiconductors perspective, Carlaw stressed the sector is a "high risk, low reward" opportunity as of now. "We are seeing a lot of price pressures even before there is any meaningful volume out there, but despite this companies such as picoChip, ST Micorelectronics, ADI and Xilinx are pretty active in the market. Most are focusing on total solutions while others such as picoChip are partnering for the radio side. Others, such as Texas Instruments, Broadcom and Freescale Semiconductor are looking on and waiting for volumes, and will either acquire the expertise or dive in and commoditize the market," Carlaw said.
The analyst's "conservative" estimate is that the semiconductor opportunity for femtocells will be about $50 million by 2008, rising to a "robust" $935 million by 2012.
Peter Claydon, one of the co-founders and COO of picoChip, which claims it is the only company shipping baseband chips specifically designed for femtocells, said "a typical chip for a femtocell costs about $10 million to design, including software." With the projected numbers for first generation femtocell baseband chips, the company would need a $2 margin per unit to recover costs, he estimated.
Claydon added picoChip is already designing a second generation femtocell chip.
ABI's Carlaw, however, suggested a more realistic design cost for a femtocell chip is in the region of $25 million to $30 million. "That is a big investment for the start-ups that are active now for a market that has yet to prove itself."