GSA Silicon Series: 4G and the Evolution of Mobile Broadband – Keynote and Panel Discussion


The latest GSA Silicon Series Luncheon: Next-Generation Networks: The Evolution of Mobile Connectivity, was held on May 4th at the Santa Clara Convention Center in Santa Clara, CA. The event started with a keynote address on converged connectivity, the rapid evolution to LTE, and the future of mobile networks. This was followed by a very spirited panel session on related topics.

Mobile broadband is today's growth engine for the telecom industry. Despite the current global economic slowdown, AT&T witnessed a 37.2 percent increase in wireless data revenue to $3.4 billion during the second quarter of 2009, adding more than 2.4 million iPhone customers. Ericsson released measured traffic data on live networks showing that data surpassed voice traffic globally in Dec 2009. The message is clear: Consumers want mobile broadband, and they want it now.

Very high expectations for 4G networks have been fueled by hype about what next-generation wireless networks will offer. While Internet access will be the most prominent 4G service, other premium services including turn-by-turn directions, VoD, pay-per-view and gaming will drive revenues and profits. 4G is also expected to enable video content to be delivered economically and enhance teleconferencing through the rapid upload and download performance. Content-driven applications have created high demand for mobile broadband from consumers. Not only are smart phones, tablet PCs, eReaders, netbooks, and handsets part of this trend, but they also fuel the need to stream content through mobile electronics creating spin-off effects. Such emerging technologies include home networking, digital TVs, medical, automotive and industrial applications. These and others are ready to move to next generation wireless networks.

1. Keynote Address: LTE and the Future of Mobile Networks, by Arpit Joshipura of Ericsson Silicon Valley


As consumers strive to achieve more connected lifestyles, mobile networks will evolve, and the mobile industry is already hard at work defining the technical solution that will allow mobile networks to meet the growing demand for wireless broadband services. The focus of this keynote will be on converged connectivity beyond phones and laptops/netbooks/ebooks and how the semiconductor industry can take an active roll in the revolution toward 50 billion connections by 2020. Today's discussions center around converged architecture, mobile data traffic explosion & 4G/LTE network deployments. The entire global ecosystem (operators, enterprises, consumers, verticals, etc.) can gain from mobility, internet and broadband coming together. Mobile network transitions (GSM to HSPA and CDMA to LTE) are surprisingly easy due to standardized, mainstream technology like LTE, IP and End to End (users to core) solutions. Mobile connectivity requires re-thinking of business values (cost of producing gigabyte of mobile data versus willingness to pay for connectivity) and the associated Mobile application ecosystem (fragmentation/verticals giving way for a horizontal interoperable layer).

Keynote Summary:

According to Ericsson, LTE is a cost-effective solution that enables the telecom industry to meet the consumer’s connectivity needs by utilizing it in smart phones, computers, Netbooks, PCs, and other mobile internet devices. LTE has been selected by 64 network operators in 31 countries. Ten LTE networks will be in service by the end of 2010 and that number could grow to 22. Up to 39 LTE networks may be in service by the end of 2012. So there was no doubt as to where Ericsson is focusing its mobile broadband efforts, initiatives and investments.

Arpit's first slide identified the key messages of his keynote:

  • Mobile Broadband is now a global revolution.
  • 4G/LTE/IP are enablers of connectivity. 3GPP & non-3GPP Migration to LTE has been surprisingly smoother than expected.
  • The next opportunity will be 50B devices with wireless connectivity by 2020, which is 10X scale of today. The entire wireless ecosystem stands to benefit from the vision of 50B connections by 2020.

check out the presentation  by clicking here.

Ericsson predicts 3.5B mobile broadband connections by 2014, vs. 400M today. The big growth areas are hand held devices and Machine to Machine (M2M) with wireless communications capability. However, the jury is out as to whether most M2M communications will need mobile broadband or much lower speed connectivity. That will depend on future M2M applications which we can't accurately predict at this time.

2009 was a watershed year for mobile broadband, with mobile data traffic eclipsing voice traffic for the first time. Mobile data traffic has been growing by a factor of 14 during the last two years. And it's forecasted to grow 100% annually over the next five years (by Cisco and others). And that's not including 3G or LTE in either China or India!

Arpit predicted, "Anything that benefits from being connected will be connected." Ericsson sees more than 6B connections by 2014, growing to 50B by 2020. The latter includes 30B connected consumer devices, based on the assumptions of 3B middle class consumers x 10 connected devices. The categories include: Personal networks, on-line homes, intelligent transport (>1B vehicles), smart utillities (>3B utility meters), industry and society (>100B embedded processors shipped cumaltively).

Ericsson sees a wireless world of "Converged All-IP Network & Services." The End-to-End All-IP Reference Architecture they envision is as follows:

click here to see the entire presentation

"Smarter pipes are enabled by IP and convergence," according to Ericsson. In particular, intelligence resides in the converged network access and edge, which supports both fixed and mobile connections. As an example, edge routers now include management, policy and charging/accounting capabilities. This, in turn, facilitates converged services that can be deployed over different types of network access.

Early deployments of mobile broadband have demonstrated the following:

a] There have been smother transitions to a newer network technology:

–Mobile Networks Transitions (GSM to HSPA and CDMA to LTE) are surprisingly easy due to standardized, mainstream technologies like LTE, IP and End to End (Users to Core) Solutions.

Comment: We think the jury is still out as to whether the transition from 3G to LTE will be smooth, or a very disruptive "fork-lift upgrade."

b] Strong Business Case: 1 Gbyte of data can be produced for 1 Euro, which permits the operator to make money on their data plans.

–Mobile Connectivity requires re-thinking of business values (cost of producing G Byte of mobile data versus subscriber willingness to pay for that connectivity)

c] Horizontalization and Interoperability is becoming a differentiator.

–Mobile app ecosystem is changing.  Fragmentation and vertical applications are giving way to a horizontal, interoperable layer (no explanation was provided as to what this new layer consists of).

Ericsson believes that 2010 will be the tipping point for real growth of built-in mobile broadband markets. They see four waves of embedding mobile broadband in devices:

a] The notebook PC: pioneered embedded mobile broadband. The notebook PC includes an embedded 3G or 3G/4G card (vs.separately sold USB dongles for notebooks).

b] The netbook PC rapid uptake; being sold like a mobile phone by network operators.

c] Emerging devices, including: eReaders, tablets,MIDs, GPS, cameras, gaming devices

d] M2M: there will be market segments with high band-width/low latency needs and others that won't produce much traffice, e.g. a sensor on a boat that signals when it's drfting away.

Arpit showed this chart depicting the semiconductor impact of the mobile broadband/Internet evolution:

›Network Packet Processing (Routers, Base Stations etc)

–Customized Performance: Power/Performance vs Cost/Flexibility

–Optimized as a System on a Chip(s)

–Latest semiconductor process (40/45nm)

›Control & Management Plane: Network Processors

›Device Semiconductors

–Scale, Cost, Size

–Part Optimized for Processing, Memory, Connectivity

Two semiconductor conclusions of this dynamic are that:

›Future (mobile broadband) services require more memory access (Mobility, P2P, Video, Personalization) and….

›Total System Optimization Required (I/O+Memory+Processing+Connectivity)

Mr. Joshipura explained why Ericsson now has such a large presence in Silicon Valley. Perhaps, the most important reason is that devices and applications have shifted (from Europe and Nokia) to Apple and Google, which have their headquarters and brain thrust in Silicon Valley (Cupertino and Mt. View, respectively). Additionally, Cisco, Intel, Oracle are also here, along with many innovative start-ups. Ericsson Silicon Valley is looking to form technology innovation partnerships here on mobile broadband and IP. Ericsson in America was said to now be the major partner to operators, fixed and mobile, services and R&D devices.

Regarding outsourcing of network operations, Ericsson is now managing 400M subscribers and providing backend network operations (billing, customer care, etc) for many network operators around the world. Sprint has outsourced both its fixed and wireless network operations management to Ericsson. In this way, "the network operators don't get bogged down in the complexity of operations," according to Arpit.

2. Panel Discussion: Beyond 3G: 4G, WiMax, LTE…What Next?

Abstract: This panel will focus on the challenges and opportunities for growth of next-generation wireless technologies.

4G networks will be one of the biggest tech-building booms of the decade, promising to take smartphones to desktop speeds, unleash new wireless devices and deliver the full potential of the mobile Internet. Conventional wisdom suggests that technology evolves when the current generation has reached its maximum usefulness. We are witnessing a skyrocketing demand for wireless data consumption, supported by a flourishing ecosystem of devices and applications, as well as a fundamental change in end-user behavior. The Internet generation is adopting a wireless lifestyle, and this appetite for data-based applications is driving the need for ever faster mobile broadband.

To meet consumer demand, two technologies hold a great deal of focus: WiMAX and LTE. The battle between WiMAX and LTE is significant and polarized. Like WiFi, WiMAX is an IEEE standard, while LTE is a 3GPP standard, as is GSM (2G), UMTS (3G) and HSPA+ (3.5G). Because of the latter, many proponents believe approximately 80% of the world’s carriers will ultimately evolve beyond 3G capabilities. However, as with all technologies, there are several obstacles to achieving this evolution. Semiconductor companies must find ways to innovate to meet operator and consumer demand. Many 3G supporters that feel that 3G still has a lot of untapped potential and that moving beyond 3G is not necessary. However, consumers want more speed, more data and more connectivity, leading to the development of next-generation networks.

Moderator: Jim Feldhan, President, Semico Research


Raouf Halim. Chief Executive Officer, Mindspeed Technologies, Inc.

Lars Johnsson, Vice President, Marketing & Business Development, Beceem Communications

Arpit Joshipura, Vice President, Strategy & Marketing Development, Ericsson Silicon Valley

Sandeep Vij, President and Chief Executive Officer, MIPS Technologies, Inc.

Richard Yeh, Director of Marketing, System LSI, Samsung Semiconductor, Inc.

Sanjeev Athalye, Director, Product Management, Qualcomm CDMA Technologies

Panel Summary:

Moderator Jim Feldhan asked the panelists, "What's driving the need for mobile broadband?"

Sanjeev Athalye of Qualcomm was first to respond: "More available spectrum is driving the transition from 3G to 4G." Wider spectrum enable today's applications to reach many more wirelessly connected devices and hence more people. It results in higher peak rates, higher average user data rates and reduced latency (which is very important for fast/short transactions). Always ON devices have also created more of a need for mobile broadband.

Raouf Halim of Mindspeed said, "It's all about video." The confluence of social networking and video content is driving the need for mobile broadband to reach more people.

Lars Johnsson of Beceem opined that 4G is based on different technologies than 3G. 4G networks (e.g. mobile WiMAX and LTE) are designed for mobile data, while 3G is not. 4G uses OFDMA (Orthogonal Frequency Division Multiple Access) as its air link, which is spectrally more efficient than 3G air links (HSPA or EVDO). MIMO (Multiple Input Multiple Output) also helps 4G – it squeezes more capacity out of the same spectrum. Users consume an average of 10G bytes of mobile data per month (I think this is what Clearwire has reported for its mobile WiMAX users in the U.S.). Lars surprised the audience and this author by stating that, "40% of the bandwidth consumed is on the uplink, and that is bad news for mobile network operators."

Sandeep Vij of MIPS said that "streaming video was a killer app." He continued, "Future apps will be based on what children are doing with social networks on laptops and these will migrate to mobile devices. Vidoe conferencing and video chat were two examples Sandeep cited. Others were getting High Definition (HD) videos from the web, rendering and uploading them to the web. As a result, 4G performance is likely to increase by 10x to 20x. The result of this performance explosion will be that the current 3G handset incumbents may not be the winners in 4G devices, which have to move large amounts of data faster and more cost effectively.

Richard Yeh of Samsung said the industry is "moving out of the monolithic world into a diverse world of customized designs and optimized solutions." As a result, semiconductor companies need a "full line-up of IC solutions and the ability to scale (presumably, IC production quantities).

Mr. Feldhan observed that medical, smart metering, and security applciations present opportunities and challenges in integrating functions, especially video, into 4G handsets. He asked the panelists to comment on this and share their experiences.

Lars said that Beceem has initiatives (with its customers) in mobile health care (including virtual physical exams), smart meters, residential home networks (where Ethernet over Power Lines is used for home distribution of data). He thinks that security will be critiically important for 4G video applications.

Sanjeev opined that most of the challenges would be related to 4G networks. This includes global spectrum fragmentation, the long lead times necessary to drive costs down and voice/data interworking for LTE. He cited the 700 MHz of upper vs lower spectrum being used by different network operators (e.g. AT&T, VZW, others) as a case in point. Operator use of different spectrum causes frequency band proliferation and skew, which is very problematic for device and semiconductor manufacturers. He also observed that it take two to three years to mature any new network technology. Wide scale adoption of a new technology require high demand, economies of scale production and lower pricing (i.e. higher volumes are needed to drive equipment and device costs down). This takes years. Finally, voice over LTE is not yet settled, but must be firm to drive mass customer adoption. This is because voice is still the key ARPU driver for most operators, while LTE (and WiMAX) are all IP networks. {Of course, mobile VoIP over LTE with 3G voice interworking could be the solution, but not many folks are endorsing that approach}.

Prompted by the longer than expected time to realize a mass market for LTE, the moderator then asked, "Will 3G and 4G co-exist for a long time?"

Sanjeev stated that cheap and pervasive (mobile broadband) data is the selling point for 3G networks. 3G device and terminal costs are starting to come down (have you noticed the dramatic price reductions for the 3G iPhone?). And 3G is just starting to take off in developing countries.

Raouf was quite realistic when he said, "New network technologies typically have a 10 year deployment cycle. As a result, 3G will dominate for the next 10 years."

Sandeep asked, "How will people use large 4G pipes? Will devices be powerful enough to cost effectively use all that bandwidth?" He noted that there are a lot of tasks going on at the same time within 4G devices. In addition, there may be virtualization, security and multiple OS's in the 4G device. With all those functions to contend with, how well will the 4G pipe be utilized? (The clear implication was that a much more powerful processor than the ARM core or AToM uP would be required in mobile devices. Of course, that's what MIPS plans to produce).

Richard was quite pragmatic in his comments, "People like HD, but the business model for HD web video (over mobile wireless networks) is not here yet. The mobile device would need to support a high frame rate, e.g. 1080p (1,080 lines of vertical resolution or 1,080 horizontal scan lines; while the letter p stands for progressive scan -meaning the image is not interlaced) at low enough power to preserve device battery life. There are similar requirements on image sensors. Also, the backend ISP capabilities need to be considered for HD video over 4G networks. The challenge will be to create one or more business models for 4G video and other applications to "survive and thrive," he said.

Mr. Feldhan then asked the panelists "can we monitor our work better (with 4G) and interject when we want to?"

Lars replied that, "Enterprise applications will drive the adoption of 4G networks." He cited mobile workers accessing 4G networks from wherever they are.

Sanjeev echoed what Arpit said during his keynote talk, "64 global operators are committed to deploying LTE." The LTE ecosystem will be bigger than the 3G ecosystem. And it will be much greater than the WiMAX ecosystem (all but the most ardent chearleaders have realized that WiMAX is just a niche mobile broadband market). TDD will be used for LTE over abundant spectrum operators now owned (note that the existing 3GPP LTE specs use FDD, because that is what was used for 3G voice). The LTE ecosystem will consist of both FDD and TD-LTE (new 3GPP specs in the works).

Lars quipped that "competitiveness drives innovation." FDD was for cellular voice, so 3GPP adopted that duplexing method for LTE. In contrast, WiMAX uses TDD. Now, operators in China and the U.S. (Sprint and Clearwire) have expressed interest in TD-LTE.  

Jim's next issue for the panelists concerned mobile device design: Will System on a Chip (SoC) be the preferred solution vs. a discrete processor (e.g. Intel's AToM) complemented by FPGAs and discrete logic?

Raouf commented that "it's all about cost and power." (We thought it might also be about performance, speed and size).

Sanjeev said "the selling point will be the integration of multiple functions into a single package." And these functions include metrics like performance, cost, board area, power consumption, etc. For example, digital baseband (MAC and PHY) and analog (RF front end, power management) technologies integrated into a single package would make for attractive and cheap smart phones possible. "Thin is still in," he said.

Richard stated that a customized solution that optimizes cost is what was needed. The exception was said to be smart phones, where it was better to decouple the processor from the baseband components. Selected RF functions should be grouped together outside of the SoC.

The first question from the audience was about mobile power consumption and battery life.

Sandeep was first to answer. He said that there is an opportunity for multi-core processors to run at the lowest possible power consumption. The concept is to turn on and off individual core processors when in use/ not in use (Of course, this in turn depends on very judicious task partitioning amongst the individual processor cores). Different levels of performance (and power consumption) can be realized per core processor. Another useful capability would be to enhance multi-threading to run on individual core processors. (Something that MIPs has evidently mastered.)

Arpit stated that choice of x86 (or Intel's AToM), ARM, or MIPS processors would be very important for realizing low overall device power consumption. 4G is more power efficient than 3G, enabling data to be delivered more quickly to the device.

Sanjeev said that Qualcomm was evaluating a video/ imaging technology that consumes power only when swtiching states.

The next question from the floor was whether there are any 4G opportunties for semiconductor start ups or "smaller players?"

Richard didn't answer the question, but noted that Samsung was "going after large markets, where its (production and manufacturing) scale is an advantage."

Sandeep said that "4G is a rapidly growing opportunity for MIPS. While MIPS processor cores are now used in the digital home, several customers are evaluating them for 4G. He cited Beceem, Altair (Israeli company) and an unnamed Asian SOC company among those customers.

Lars joked with the audience, "Don't mess with Beceem!" That is, don't threaten their dominance of the WiMAX chip market (Beceem recently announced they'd also be supporting LTE on the same chip). Then he identified several functions that could be effectively be done by smaller companies: RF front ends, analog switches, adjustable filters/tuners, sensors, timers.

Raouf cautioned the would- be 4G IC players to "be prepared for a long, expensive haul." Analog, RF, baseband designs all involve big investments and may take twice as long to realize as the new company might expect.

Sanjeev summed up by saying that "the majority of silicion in 4G devices will be primarily from large chip companies."  (But we observe that start-up Beceem is the undisputed leader in shipping WiMAX chips and chip sets).

0 thoughts on “GSA Silicon Series: 4G and the Evolution of Mobile Broadband – Keynote and Panel Discussion

  1. Author's Opinion:  Mobile Broadband Market for Semiconductor Companies
    Regarding semiconductor content for mobile broadband, I believe that the large equipment companies will prefer their own ASIC/FPGA designs over merchant market silicon.  The device market will be primarily SOCs for digital functions, with some mixed signal technologies used in air interface ULSI components.  The real battle will be the processor(s) selected for the various mobile broadband devices/ tablets/ gadgets.  ARM core has the overwhelming lead now.  But in the future, Intel's next generation (lower power consumption) AToM processor is a contender as is MIPS multi-core (for SOCs) and proprietary processors (like Apple is using in the iPhone and iPad.  I believe that's where the real semiconductor battle will be for the 3G+ and 4G mobile broadband markets.

  2. The bandwidth chart from Ericsson is indicative of the value that people place on getting content where and when they want.  The question is whether wireless will have to part of a Communications Service Providers' whole product strategy in order to remain competitive.  It also points to the FCC's concern about opening up spectrum for mobile applications.   

  3. The wireless carriers are between a rock and a hard place, IMHO.  For years, they have counted on adding more subs and locking them up with a multi-year contract that subsidizes the smart phone that they sell.
    Well, it looks like that game is over!  Additional subscriber growth is negligile and more and more subs are opting for pre-paid plans from lower cost carriers like Metro PCS.  In addition, the two largest carriers (ATT and VZW) that are selling popular smart phones have a huge challenge in preventing their networks from becoming saturated with all the data and video traffic they have to handle.  I don't know how they can make money off mobile data services, in light of these issues and market dynamics.

  4. Following up on my last comment, there are two ways for mobile data carriers to make money off smart phones and similar gadgets:
    1.  Early termination fees (we've previously commented about that)
    2.  Tiered pricing for mobile data users.  Here is what AT&T just announced for that:
    AT&T Mobility (NYSE:T) this morning made the nation's first definitive step toward tiered data pricing by instituting usage limits on what were previously unlimited smartphone data plans. Executives from the nation's largest carriers have long discussed a move toward tiered, bucketed data pricing scenarios for smartphones, but AT&T is the first major U.S. carrier to implement the changes.
    Perhaps not surprisingly, the changes will go into effect June 7–the same day Apple (NASDAQ:AAPL) CEO Steve Jobs is widely expected to unveil the company's next iPhone.
    Here's how AT&T's new plans break down:

    DataPlus: Provides 200 MB of data for $15 per month. If customers exceed 200 MB, they can purchase an additional 200 MB for $15. AT&T said 65 percent of its smartphone customers use less than 200 MB of data per month on average.
    DataPro: Provides 2 GB of data for $25 per month. If customers exceed 2 GB, they can purchase an additional 1 GB of data for $10. AT&T said 98 percent of its smartphone customers use less than 2 GB of data a month on average.
    Tethering: Smartphone customers–including iPhone customers–with the DataPro plan will have the option to add tethering for an additional $20 per month. Tethering for iPhones will be available when Apple releases iPhone OS 4 this summer, AT&T said.

    The plans stand as a stark refutation of AT&T's previous pricing–which were in line with much of the rest of the domestic and international wireless industry–offering unlimited data access for a flat $30 per month. That the carrier has reversed course likely is a reflection of the dramatic increases in data traffic on AT&T's network, primarily generated by iPhone users. AT&T's new data plans serve the dual purpose of preventing excessive use while rewarding light data users with cheaper prices.
    "While the convenience, simplicity and peace of mind that comes with an unlimited plan helps to drive adoption and reduces customer care costs, it is also unsustainable," wrote analyst Roger Entner of The Nielsen Company. "It certainly has driven adoption through predictable pricing, but with a long-term downside risk to the overall business model and the financial viability of the entire industry. … By lowering the data component of smartphone plans, among them the iPhone, they [AT&T] are stimulating demand, just like the shift from unsubsidized iPhone pricing to subsidized iPhone pricing drove adoption."

  5. 1. Of the forecast 50B (by year 2020) devices connected by wireless, a point was made about not all devices needing broadband, something I agree with.  How ever, the 50B number seems to be vastly underestimated when you look at HP's CeNSE initiative (central nervous system for earth) which proposes to use one trillion devices. Many of these will be mesh-networked sensor nodes with much lower bandwidth needs but will need to be connected nonetheless.

    2. Even with 50B connected devices by 2020, what about the power source challenges? Are there enough resources for making batteries to store energy for that many devices? The existing battery technologies have largely lagged behind their counterparts in electronic components and systems. I believe there needs to be many quanta-leaps in this area to making connectivity for that many devices a reality.

    3. Adoption in developing countries: an overwhelming majority of users in Asia do texting much more than voice & data and it is not just the younger crowd.  About 75% of the mobile users use some form of SMS/TMS in the world today & this number is even higher if you look at Asia separately. In that market, it is predominantly text, not text with video & images. It wasn't clear from the presentation how these users would be motivated to upgrade to 4G when their needs are met today with 3G?

Leave a Reply

Your email address will not be published. Required fields are marked *

I accept that my given data and my IP address is sent to a server in the USA only for the purpose of spam prevention through the Akismet program.More information on Akismet and GDPR.

This site uses Akismet to reduce spam. Learn how your comment data is processed.