In the first instalment of Wireless Watch’s Mobile World Congress special edition, we looked at a selection of operator trials of pre-5G networks and services. Now we turn our attention to the vendors’ roadmaps, and the most interesting insights they provided in Barcelona, into developments which may well find their way into future standard or de facto technologies.
Nokia was the most aggressive of the big names in claiming to be 5G-ready – although it was a heavily overused term throughout the show. The Finnish vendor went a step further and said operators would be able to offer ‘5G’ services as early as 2017 using its new AirScale RAN platform (clearly angling for those operators which are promising ever-earlier ‘5G’ services, mainly around major sporting events in Russia and Korea, but which will, of necessity, have to use pre-standard or heavily customized kit).
AirScale will accelerate the transition to 5G when it comes by providing a smooth migration path from current LTE, Nokia promised, rehashing one of the most popular messages which vendors relay when a new generation of technology is looming. Somehow, when that upgrade becomes necessary, it’s never as automatic, low cost and software-based as the roadmaps suggest – virtualized flexible cores and software-defined radios will help, but there is still the question of the antennas, for instance. But Nokia, like its rivals, needs to persuade operators to keep investing in 4G now, rather than wait for 5G over the horizon.
CTO Hossein Moiin said the AirScale demo was “the industry’s first demonstration of how 5G will work in practice, going beyond previous experimental systems”, making 5G “no longer a distant vision” – ambitious stuff, considering the work on standards has only just begun. But some crucial elements of 5G are already accepted, Nokia argues, and these are the focus of AirScale – in particular, ultra-low latency, and a system designed with machines as well as humans in mind.
AirScale’s claim to be a vision of the future rests on its cloud-based architecture and open interfaces, which allow it to support all the radio technologies simultaneously in one base station; and to use any architecture topology with huge levels of scalability, all defined in software from the cloud. It also claims to use 60% less energy than Nokia’s current platforms. The server end of its Cloud-RAN solution uses Nokia’s AirFrame IT hardware, launched at MWC 2015, and Mobile Edge Computing is implemented on those same servers in order to harness information from the radio effectively and support low latency services.
Baseband units can be chained to support massive capacity and IoT connectivity, and a step-by-step road to virtualization is supported by Nokia’s NetAct and CloudBand offerings, which can work with legacy RANs and C-RAN at once. As well as the multiband AirScale base station, the vendor added AirScale WiFi (access points plus a WiFi controller, also running on AirFrame), and a common software layer which runs across small cells, macrocells and WiFi.
CEO Rajeev Suri said: “I believe that 5G not only must happen faster than expected – it will happen faster than expected”, which was why vendors could not wait for standards to be finalized before launching 5G products – a line also taken by ZTE with its Pre5G portfolio, though not by Huawei, which prefers the term 4.5G for its most advanced network offerings, and argues that these will meet operators’ needs for years to come, certainly until fully standardized 5G can be launched.
“5G is different. 5G must happen fast because important use cases demand it. If we know that 5G can help save lives, improve our environment and make our lives better, we need to move faster,” Suri said.
ZTE has unveiled the latest developments in its Pre5G range, the Ultra Dense Network (UDN), claiming this can provide a highly functional bridge to emerging 5G platforms. The UDN works with existing networking models but aims to enhance the user experience with a combination of techniques including interference management, suppression and mobility enhancement.
The system adopts system frequency multiplexing to address interference caused by
overlapping cells in dense networks. According to ZTE, this technique can boost downlink rates by 10 times in areas of cell overlap and will be an important component of 5G.
The equipment also uses other expected 5G technologies, but on current network infrastrucfture and sites – like Nokia, holding out the hope of a smooth migration down the road. These include massive MIMO and Multi-User Shared Access (MUSA), which ZTE has been testing for over a year. Early in 2015, it announced the first commercial base station to implement its pre5G architecture, particularly massive MIMO and MUSA, and this has completed field tests with a number of global operators, including China Mobile and Softbank.
Zhang Jianguo, wireless general manager at ZTE, said Massive MIMO “will be in commercial use in China and in other countries later this year”.
Qualcomm and Ericsson:
Qualcomm and Ericsson have announced their latest collaboration, and will work together on early trials and verification of key 5G technology components, to support the technical work required for 3GPP standardization in Release 15. The companies also said they would drive interoperability in alignment with 3GPP to enable rapid adoption of new 5G standards.
“As we did in both 3G and 4G, we are excited about collaborating with leading operators and industry stakeholders such as Ericsson in the development of a unified, more capable 5G platform” said Matt Grob, Qualcomm’s CTO. “Now that the vision and interest for 5G are well established, it is time to focus on the technical and engineering work required to support operator trials and commercial network launches.”
Intel was determined to put paid to any perception that it might be an outsider in the 5G race. Not only was it focusing heavily on virtualization and Cloud-RAN – changes to the mobile network where it has the most obvious opportunity to enter the inner circle – but it was talking up partnerships with Ericsson, Nokia, LG, Cisco and Verizon to develop technologies for the IoT, smart cities, driverless vehicles and augmented reality – all important drivers of 5G development.
“5G represents a significant shift for these networks and we think it’s essential to get ready ahead of the curve,” said Aicha Evans, general manager of the Intel communication and services group. “Rather than just being about a personal computing platform, it’s about everything that computing can connect and how it connects.”
Among the specific announcements were IoT-oriented connectivity solutions like the
Atom x3-M7272 wireless communication platform for “automotive applications capable of powering advanced security features”; and the XMM 7120M LTE modem, optimized for machine-to-machine applications.
Intel also announced a new collaboration with Cisco and its strategic partner Ericsson, to develop a 5G router. This will be submitted to the Verizon 5G Technology Forum, of which all there vendors are members, and will aim to improve speed, latency and IoT scalability for business and residential customers.
NEC has developed a prototype of a compact, A4-sized massive-element Active Antenna System (AAS) for 5G small cells operating in the lower SHF (super high frequency) bands, from 3 GHz to 6 GHz.
As well as its small form factor, the combined antenna/RF solution boasts fully digitized antenna beam control and MIMO pre-coding to improve the precision of beamforming and increase. The vendor said that, used in conjunction with NEC’s spatial multiplexing technologies, the AAS can achieve more than 10 times greater per-cell throughput than conventional LTE base stations. It is now working on trials of the AAS with NTT Docomo.
“5G communications achieve higher speeds and increased capacity through the utilization of high frequency bands which are capable of securing wider bandwidth. On the other hand, high frequency bands face the issue of heavy propagation loss in communications. As a solution to this issue, NEC has focused on beamforming technologies which improve communication distances and reduce interference,” said Nozomu Watanabe, general manager of the firm’s mobile RAN division.
NEC also issued three white papers related to 5G – ‘Optimum Network Architecture for Full-scale IoT’; ‘NFV C-RAN for efficient RAN resource allocation’; and ‘Massive MIMO for High-capacity Mobile Access’. These are the three areas which the company believes will be most central to 5G’s architecture and business case.
For the IoT, it considers coordination between Mobile Edge Computing and the cloud as well as full context awareness when allocating resources. The C-RAN paper outlines NEC’s approach to this architecture, including its proprietatry inter-cell interference mitigation techniques and its fronthaul design, and describes a migration to 5G based on “virtualization of cells”. And the third paper shows how Massive MIMO can be applied to small cells, using technologies like NEC’s new AAS.
NEC is also working with MediaTek to develop 5G air interface and chipsets.
This article is part of our Wireless Watch subscription service. Learn More
Qualcomm, free from the looming shadow of its strategic review, has wasted no time in gathering its weapons for a major attack on new markets in 2016. The Consumer Electronics Show saw it unveiling its connected home and car efforts, and this week it has followed up with a Chinese joint venture to boost its nascent efforts in the server market.
This also reflects the rising importance of China in its strategy, not just as a huge and competitive market, but a source of new partnerships such as this new one with Guizhou Province, or its strategic alliance with foundry SMIC. Diversifying its foundries in order to guarantee supply of key components and harness the most advanced processes is critical to any fabless chip vendor, of course, and Qualcomm has also added Samsung to a base which was once the sole preserve of Taiwan’s TSMC.
Intel, too, sees 2016 as a critical year to establish its power in the new markets it has earmarked for growth, and has its own strategic relationships in China to help it achieve that, notably with Rockchip and Spreadtrum for the mobile and embedded sectors. It also talked up its Internet of Things roadmap at CES, and then followed up, this week, with a deeper focus on infrastructure platforms.
Rethink Wireless is a taster of its parent publication Wireless Watch. Click the following link to read the full story by signing up for a 4 week FREE trial. Wireless Watch Trial
Rethink Wireless is a sample of our full paid service Wireless Watch, click the link below for 4 week FREE trial. Wireless Watch Trial
With Qualcomm conducting a strategic review, under pressure from activist investor Jana Partners to break itself in two, there is plenty of speculation about the likely outcomes.
Some believe that, if the company does split its licensing and chip technology businesses (see Wireless Watch August 23 2015), the latter would become prey for an acquirer, probably Intel, and so continue the ongoing wave of M&A among semiconductor firms.
Reuters calculates that the Qualcomm chip business would be valued between $30bn and $40bn and could attract Intel as a way, finally, to achieve a strong position in the wireless market. Other analysts speculate that the division could be a target for Samsung, or for a Chinese group (despite the likely US government blocks).
But it remains true that, without the licensing arm, the chip unit is far less valuable than it currently is. It would be deprived of the profits from the patents royalties, which help to fund Qualcomm's massive investment in R&D, which in turn helps the chips to remain at the cutting edge.
Qualcomm itself is thinking in terms of being the acquirer, not the prey. CEO Steve Mollenkopf told Bloomberg: "Qualcomm is very likely to be some form of actor in the consolidation of the semiconductor industry … The timing of which is always the debate." The company will look at acquisition opportunities as part of its strategic review process, he said.
While not identifying possible targets, he said Qualcomm wants to expand outside its traditional handset sector into growth markets like connected cars and healthcare. It has already made substantial developments in these areas but could accelerate the ROI with an acquisition.
He also noted that the firm is better placed to make a major purchase now it has settled its antitrust lawsuits in China, bringing greater predictability to its revenues there. "It's an interesting time to be a scale player in the semiconductor industry, particularly from a position of strength," Mollenkopf said. "We've had strategic flexibility, but it's important that you have your home business well structured for the future first."
Among recent chip acquisitions are Intel's of Altera, NXP's of Freescale, and Avago's of Broadcom, leaving limited options in the top tier for Qualcomm. Nvidia, or even the combined NXP/Freescale, might be targets, speculate Wall Street analysts.
BY CAROLINE GABRIEL
Around the turn of the year, Samsung was forced to make some bold statements of direction as the harbingers of doom leapt on the declining growth and market share in its most important business, smartphones. It indicated that it would put more funds and focus on other revenue streams, particularly chips and displays, and the former strategy already looks to be having some effect.
The mainstay of Samsung's chip operations is memory, but it also aims to grow its mobile processor and modem activities, and its foundry business. Reports that it will use its own Exynos SoC in more Galaxy S6 models than previously expected (at the expense of Qualcomm) shows the company becoming more self-sufficient, though it also needs to sell the product to more third parties to challenge the leaders, especially if its own smartphone demands fall.
Other reports said that Apple, hard as it has tried to cut its ties to its biggest smartphone rival, is swinging back towards Samsung as the primary manufacturer of its mobile processors. Sources say the latest model, the A9, will mainly be made by Samsung in Korea, and also by its partner GlobalFoundries, a blow to TSMC, which was the main foundry for the A8.
Improving sales of chips and displays are already affecting quarterly figures. According to Samsung's preliminary statement of its first quarter results, those two businesses saw growth, even while the mobile unit continued to contract, leading to an anticipated 30% year-on-year drop in Q1 profits - the sixth quarter of profits reduction in a row (though it was better than analyst estimates). Operating profit for the quarter will be around KRW5.9 trillion ($5.4bn), said Samsung, while revenue should fall by 12% to KRW47 trillion. Full results will be reported at the end of April.
Beneath the depressing headlines, the semiconductor unit will post a profit this year, analysts expect, and in Q1, it is predicted to make sales of 11.1 trillion with operating earnings of KRW2.7 trillion. The display division is expected to report operating profit of KRW450bn, boosted by rising demand for ultra-thin displays for affordable smartphones. The higher end OLED screens are mainly used in Samsung's own high end models.
Operating profit at the consumer electronics division, which includes TVs and appliances, is expected to fall, as are earnings at the mobile unit - likely to post operating income of KRW2.3 trillion on sales of KRW27.1 trillion, according to an analyst poll by Bloomberg. Daewoo Securities said Samsung shipped 81m smartphones in the quarter, up 7% from Q414, and its current quarter should be boosted by the shipment of the Galaxy S6 and S6 Edge from April 10.
By Maury Wood, Principal Analyst
Wi-Fi infrastructure has been evolving in two ways, in terms of performance and of usability. A significant wave of hotspot upgrades is under way, driven by the uptake of 802.11ac and of Hotspot 2.0/Next Generation Hotspot. These two technology shifts are the most important in the shift of public Wi-Fi from best effort towards carrier-grade.
2x2 MIMO 802.11ac is now a standard feature in new smartphones, phablets, tablets and notebook computers. By early 2015, new iOS and Android radio firmware will enable Multi User MIMO functionality in smartphones, potentially doubling Wi-Fi throughput as 11ac Wave 2 Access Points are deployed. This article highlights some of the key findings of the latest and unique report entitled “Wi-Fi Blasts Ahead with 11ac Wave 2 Chipsets” published this week by EJL Wireless Research, a research partner of Maravedis.
Promise of Gigabit Wi-Fi
Consumer client devices and access points (both consumer broadband gateways and enterprise-class / carrier-class) using 11ac Wave 2 technology promise nearly instantaneous (very low latency) data transfers with high Quality of Experience / Quality of Service. They will be capable of supporting multiple ultra high definition streaming video flows, including UHD “telepresence” video conferencing flows, as well as hundreds of clients per AP. Additionally, 11ac Wave 2 client devices and access points will offer increasingly robust and comprehensive wireless plus wired security as well as content-aware traffic engineering features.
Complex System Design Risks
Semiconductor companies who can supply both Wi-Fi radios and companion digital System-on-Chip (SoC) processors plus the software and system reference designs are at an enormous advantage over microchip companies who can only supply the digital SoC processor. Wi-Fi systems, such as access points and wireless broadband gateways, are extremely complex hardware (requiring digital, RF and mixed-signal integrated circuits) and embedded software systems. The overwhelming majority of WLAN equipment OEMs are uninterested in attempting to integrate these complex systems using radios and processors from different suppliers. A good illustration of this complexity is the Netgear R7500 Nighthawk X4 consumer WLAN router, which integrates the Qualcomm IPQ8064 with the Quantenna QSR1000 radio, and despite this combination of “best-in-class” silicon components, does not achieve performance that is superior to Broadcom’s chipset (radios and networking processor).
Integrated System Solutions Rule The Roost
Embedded software is frequently the critical determinant of Wi-Fi system performance, and suppliers of both radios and companion processors deliver optimized system software, as well as comprehensive system-level application / software integration engineering support to their customers. For this key reason, this report does not include analysis of general purpose embedded processors from AppliedMicro, Cavium, Freescale (including the Comcerto C2200 and QorIQ T1020), and LSI Logic, none of whom offer companion Wi-Fi radio products.
Super Smart Access Points Plus Storage
The purpose-built Wi-Fi access point networking processors most recently announced by Broadcom, Qualcomm and Marvell have enormously powerful general purpose multi-core ARM CPUs and specialized accelerator engines, and (concurrent with Wave 2 radio performance enhancements) herald a new era ofsuper-smart access points able to run multi-Gbps line rate unified threat management, adaptive / intelligent QoS, continuous spectral analysis and cognitive RF interference avoidance, deep packet inspection and other advanced embedded applications. This order-of-magnitude lift in access point compute capacity is occurring simultaneously with the trend towards controller-less cloud-managed enterprise AP architectures, as well as the trends toward Network Function Virtualization (NFV) and Software Defined Networking (SDN).
Broadcom – Market Leader
Among the top 30 best performing (using maximum 5 GHz downlink throughput as the metric) 11ac Wave 1 and early Wave 2 WLAN routers, using test results from Small Net Builder (www.smallnetbuilder.com), Broadcom chipsets occupies 22 slots (including 8 of the top 10), Qualcomm occupies 7 slots, and Marvell occupies 1 slot in results posted in October 2014. We believe that Broadcom is the performance leader in Wave 1 Wi-Fi chipsets today.
With more than 100 known 802.11ac production design wins, Broadcom is the current leading supplier of Wave 1 802.11ac chipsets (radios and companion networking processors), but Qualcomm (with 58 known 802.11ac Wave 1 production design wins) is now poised to threaten Broadcom’s leadership market share in 2H15 and 1H16 with strong recent Wave 2 product announcements. Broadcom is currently under competitive pressure to announce 11ac Wave 2 radios and upgraded fully carrier-class StrataGX processors.
Wave 2.5 Radios and RF PAs in 2H15
Announced Wave 2 11ac radios (from Qualcomm and Quantenna) do not have support for 160 MHz or 80 MHz + 80 MHz bonded channels. EJL Wireless Research terms next generation 11ac radios with 160 MHz bonded channel support as “Wave 2.5”, and forecasts this capability will appear in 11ac radios announced in 2H15. As the FCC, ETSI and other international regulatory bodies open up additional 5 GHz spectrum for Wi-Fi use, this will become a more important feature (potentially doubling maximum PHY data rate). Companion RF front-end ICs with adequate linearity performance to support 160 MHz channels are likely to be announced from suppliers such as SkyWorks and Qorvo (RFMD / TriQuint) in that timeframe. This report also compares the cutting-edge infrastructure radios and Wi-Fi networking SoC processors from the nine suppliers covered.
Download the brochure of this must-have report.
Chip giant beats Wall Street forecasts with quarterly results, but warns of licensing shortfalls in China
By Caroline Gabriel
The figures comfortably exceeded Wall Street estimates - analysts had targeted non-GAAP earnings of $1.22 a share on revenue of $6.52bn. However, this was only the second time since 2010 that Qualcomm had reported less than 10% year-on-year revenue growth in any quarter, a sign of the growing competition in its core smartphone processor space.
The cycle of warnings followed by over-achievement is not, in this case, just a corporate tendency to manage Wall Street expectations but a genuine sign of the unpredictability of the Chinese market, on which - like most companies in the mobile food chain - Qualcomm is dangerously reliant. "The company's guidance is becoming harder to achieve given the apparent delays in the roll-out of LTE in China," Bill Kreher, an analyst at Edward Jones & Co, told Bloomberg in April. "In the near term, results may be choppy."
His predictions are proving right, and Qualcomm CEO Steve Mollenkopf admits: "The launch of LTE in China is very important to Qualcomm, and it's difficult to predict." The company is particularly eager to see China Mobile's base convert to 4G, as it has created a TD-LTE iPhone for the carrier, but also because it has effectively been excluded from royalty revenues in the operator's 3G technology, TD-SCDMA.
There are other issues in China however, including probes by the NDRC government body into antitrust and corruption - seen by some as part of a broader assault on the US firm's market and IPR position - and difficulties in getting full revenues out of the complex customer base. In its new quarterly statement, Qualcomm reduced its outlook for the current fiscal Q4, because some Chinese licensees "are not fully complying with their contractual obligations to report their sales of licensed products to us".
It cited "certain licensees under-reporting a portion of their 3G/4G device sales and a dispute with a licensee" as well as possible delays in signing new licences while the NDRC investigation is ongoing.
The gap is significant - while Qualcomm expects 1.3bn 3G/4G devices to ship in calendar 2014, the number that will be reported to it for licensing purposes will be between 1.04bn and 1.13bn, because of "units that we believe may not be reported to us, are in dispute or are currently unlicensed. We are taking steps to address these issues."
Qualcomm president Derek Aberle said on the analyst call that "we are experiencing some near-term challenges in the licensing business, particularly related to China. This is something that we will take care of. "But he added, echoing other unknowables in the Chinese business, "the timing is pretty uncertain".
Licences deliver the bulk (at least two-thirds) of Qualcomm's profits and shortfalls in that area cannot be fully offset by any strong trends in chip sales, which have far lower margins. So the disputes will hit the current quarter - net income in the quarter ending in September will be $1.03 to $1.18 a share, Qualcomm forecast, disappointing analysts, who had looked for $1.23.
However, chips account for most of the firm's revenues and rising sales of high end chips will propel revenues in fiscal Q4 to between $6.5bn and $7.4bn, in line with consensus Wall Street predictions of $7.13bn. In fiscal Q3, Qualcomm shipped a record 225m chips, up 31% on the year-ago period, and the number could rise as high as 245m in the current quarter, which would be a 29% increase. New launches from its two largest customers, Apple and Samsung, as well as Chinese LTE, should boost the second half of the year, though there are fears of those handset giants losing share to lower cost vendors, which may not be Qualcomm customers - especially Chinese manufacturers, which often turn to local suppliers such as MediaTek.
Those competitive shifts, and a general move towards lower cost smartphones, are challenging to Qualcomm and will accelerate its efforts to expand in other markets such as the WiFi home and the internet of things. But those trends are fairly well understood and have been factored into analysis of the US giant for some time. By contrast, the Chinese market, especially the licensing and antitrust issues, are creating nervousness with their unpredictability.
Suji De Silva, an analyst at Topeka Capital Markets, told Bloomberg: "Qualcomm told investors that they had the China customers under contract, that it's all worked out. Now we're getting a sense that it's still a challenge. Qualcomm is trying to monetize its technology and China is a more challenging market."
Qualcomm has been investing heavily in China to build a local ecosystem and strengthen its ties with the big three operators there. Simultaneously with its results statement, it announced a commitment to plow up to $150m into Chinese start-ups at various stages, in the important growth areas of ecommerce, semiconductors, education and health. The activity will be managed by Qualcomm Ventures and the first recipients of funding are Cambridge WoWo and Boohee, in mobile education and healthcare respectively.
The vendor was keen to stress its long term engagement in the Chinese market and said it has had several Chinese investments with successful exits (such as Enorbus (acquired by Walt Disney); Aicent (acquired by TA Associates); and NetQin. Other investments in the country include rising handset star Xiaomi, as well as Thundersoft, MadHouse, CooTek, Yongche, Dolphin Browser, Alo7 and Hawkeye.
With a clear nod to the antitrust investigators, Mollenkopf said in a statement: "Since first introducing our technology and products in China well over a decade ago, Qualcomm has contributed to China's wireless industry through investing in research and development, licensing our advanced technologies, and providing the most advanced chipsets to Chinese companies. Our strategic collaboration with and technical support of the Chinese wireless industry has helped this vibrant ecosystem, helped drive direct and indirect employment, and contributed to economic growth in the entire Chinese wireless industry."
UK firm reports seasonal slowdown, but licensing revenues up by 42%, pointing to stronger growth ahead
ARM's revenues were up 17% year-on-year to $309.6m (though top line growth in UK pounds was 9%, to £187.1m), which was solid for the time of year, though slower
The UK company said its adjusted operating margin improved to 48.9%, from 48.6% in the year-ago period and earnings per share were 3.91p, from 0.75p. The results included an £8.4m restructuring charge connected to layoffs of 130 staff.
The biggest contributor to revenue growth was licensing, up 42% in dollar terms to $146.1m. ARM signed 41 new licences during the quarter. Overall shipments also rose in the quarter, up 11% year-on-year to 2.7bn chips.
By contrast though, royalty revenue grew by only 2% in dollar terms, and the company blamed slowing demand for smartphones because many operators are trying to sell off 3G models while transitioning to 4G. That has created some slowdown at the high end, amid a general rebalancing of the handset industry's growth towards emerging markets and lower cost smartphones.
CEO Simon Segars acknowledged, on the analyst call, that royalties had been hit by "seasonal trends in inventory management in parts of the electronics supply chain" as carriers worked through their 3G inventory. However, he believes this trend will create a shift towards higher end and 4G models in the second half of the year and pointed to a "healthy pipeline of opportunities", which should boost revenue growth in the third and fourth quarters.
"Our continued strong licensing performance reflects the intent of existing and new customers to base more of their future products on ARM technology," Segars said. "This bodes well for growth in ARM's medium and long term royalty revenues."
However, many analysts are cautious about ARM because of its exposure to smartphones, a segment in which prices and growth are slowing. Jasmeet Chadha of Bernstein Research told the Financial Times: "Long term, the view across the market has been that smartphone growth will slow. Half of ARM's business is exposed to smartphones. There are opportunities in the internet of things and wearables, but these will ramp over several years and are small today and do not yet represent half of ARM's business. Structurally, the group's royalty growth is likely to slow." The rising threat of Intel in core markets, and the strength of the British pound, are also negatives around ARM at the moment.
But the firm has been investing huge efforts in diversifying its business and building on its dominance in handsets. It has released processor designs, and announcing licensing deals, in several new areas. Its first 64-bit platform is targeting high performance devices and servers, while it has been expanding its microcontroller and ultra-low power architectures for the internet of things.
The biggest bellwether of the mobile processor market, Qualcomm, reports its quarterly results on Wednesday. Like ARM, it expects the second half of the year to be significantly stronger than the first as it deals with falling handset average selling prices and the 3G-4G transition in China and other key markets.
Q2 results show resilience in PCs, growth in Chromebooks, but continuing struggle to turn mobile processors into profits
In Intel's second quarter financial report, mobile losses continued to mount, but PC revenues held up surprisingly well. Overall, the company reported record quarterly microprocessor shipments, and added $1bn to its year-ago revenue figure, turning in $13.8bn for the quarter. Operating income was $3.8bn, net income $2.8bn and earnings per share 55 cents.
Intel did not give exact processor shipment numbers, but credited that business with its strong figures, and particularly singled out Chromebooks and its low power Bay Trail system-on-chip. Slow to get the power-efficient SoC right in the past, Intel was in danger of ceding not just the handset market, but all the emerging new post-PC form factors, to ARM-based rivals. However, Bay Trail and other upcoming SoCs have finally given the company an offering which can genuinely go up against ARM in the low power game.
"With the ramp of our Bay Trail SoC family, we have expanded into new segments such as Chrome-based systems, and we are on track to meet our 40m unit tablet goal," CEO Brian Krzanich said in a statement. "In addition, we hit an important qualification milestone for our upcoming 14nm Broadwell processor, and expect the first systems to be on shelves during the holidays." The 14nm shift, together with 3D transistors, will put Intel ahead of the field in terms of process, indicating the advantages of its control of its own production, and enabling it to reduce power, size and cost of chips further.
Other upcoming chips for the year include Cherry Trail for tablets, which will ship by the end of the year, and Broxton, the next generation low-power Atom chip. Next year, Intel will release Skylake, for PCs and tablets.
However, the mobile and communications group continues to struggle. Its revenue was down 67% on Q114, and 83% year-on-year, to just $51m, a drop in the ocean for a company which has often acknowledged that mobile expansion is critical to offset the decline of the PC. The group lost $1.1bn in the quarter and has lost $2bn already this year.
Intel has come very late to the smartphone party, and despite some recent design wins and a much improved offering, it still has important disadvantages compared to Qualcomm, notably the lack of a fully integrated processor/modem solution for LTE - essential for low cost, low power handsets.
CFO Stacy Smith recognized the urgency of the situation, saying that, if Intel cannot integrate mobile connectivity into communications within 3-4 years, it will be locked out of critical segments of the market. Many would say Smith was being optimistic on the timelines.
Krzanich said on the analyst call: "I believe, over time, we can make this a profitable business. We have some ground to make up in moving LTE to Category 6 and having our products designed for this segment. As we turn into SoFIA [the all-in-one SoC family], you'll see a family of product really targeting this space and that's how we become profitable."
Google's Nest unit, Samsung and ARM head up latest IoT-focused standards group, with focus on 802.15.4 wireless home networks
The Thread Group initially contains Nest, the smart home gadgets maker owned by Google, and Samsung, along with ARM, Freescale, Silicon Labs, Yale Security and ceiling fan maker, Big Ass Fans. Samsung and Google both have devices as their entry point to the smart home, and from there the broader IoT, but have ambitions to influence the whole stack, using 'open' vehicles to try squeeze mutual arch-rival Apple back behind its garden walls.
In contrast to some other recent IoT groupings - such as the AllSeen Alliance, based on Qualcomm's AllJoyn technology, and the Open Interconnect Consortium, led by Intel - Thread aims to standardize the physical network which could then support any of those higher layer standards.
Thread is initially heavily focused on 6LoWPAN, because it is already used by Nest, and because it supports IPv6, important to ensure the IoT is future-proofed against running out of address space. 6LoWPAN is effectively a version of IP for the embedded space, providing a compression format for IPv6 that is optimized for low power, low bandwidth wireless links.
But the new body also hopes to lure the larger base of ZigBee developers, claiming many ZigBee devices could be upgraded to support Thread with just a software update. Attracting a home-focused ZigBee company like GreenPeak would be a valuable endorsement in the first major target market, the smart house.
Thread will add software to the 802.15.4/IPv6 foundation, for functions such as routing, set-up, security and device wake-up, to standardize these capabilities and reduce power. The Thread group will provide testing and certification for its specifications, emulating WiFi and Bluetooth rather than the more splintered ZigBee. Some Nest products already use an early form of Thread, rather than vanilla 6LoWPAN, pointing to the heavy influence of Google's subsidiary on the shape of these specs, though there is also likely to be considerable input from ARM via its Sensinode acquisition. The Finnish software firm was a significant contributor to 6LoWPAN and other low power M2M standards.
To win sufficient critical mass to become a de facto standard, Thread will need to prove superiority over Bluetooth Smart. The new group's backers argue it will do this because 802.15.4 supports true mesh - a useful architecture for home networks and not currently enabled by Bluetooth - and the new Thread additions will promise stronger encryption and IPv6, and even lower power consumption.
The appeal for ZigBee stalwarts - which include Samsung, the only major handset maker to propose implementing the protocol in mainstream smartphones - is that the industry weight of Google might give their standard a boost in the home market, where it has had far less impact than in the industrial world.
"Existing wireless networking approaches were introduced long before the IoT gained ground. The Thread protocol takes existing technologies and combines the best parts of each to provide a better way to connect products in the home," said Vint Cerf, VP and chief internet evangelist for Google and advisor to the Thread Group.
Provides vendor neutral reference board to help smaller developers accelerate their moves to 64-bit platforms
ARM's 64-bit architecture, ARMv8, is gaining rapid momentum. Some licensees, such as Qualcomm, will do their own designs and Android implementations, but Google's operating system is all about extending to all levels of the chip industry. Linaro's recently announced 64-bit reference board for developers, Juno, targets those smaller providers, and aims to encourage proliferation of 64-bit ARM/Android devices. Now it has followed that up with a full port of the open source Android code (AOSP) to ARMv8-A.
This is part of Linaro's 14.06 release and, together with the ARM Development Platform (ADP), aims to accelerate 64-bit Android availability on silicon. James McNiven, general manager of systems and software at ARM, told EETimes: "Our collaboration with Linaro will enable our partners to create devices that will drive the best next generation mobile experience on 64-bit Android operating systems, while also providing full compatibility with today's 32-bit mobile ecosystem that is optimized on ARM-v7A."
Linaro, established in 2010, seeks to create a unified Linux code base for the ARM platform, with the related goal of fending off Intel, which hurls fragmentation as one of its most powerful missiles against its chief architecture rival in the mobile world.