His observation that the new MSi3113 chip is a front end to SDR (software-defined radio or software modem) with the demodulation implemented in software on the host processor is right and the key is whether the compute resource is sufficient for the application. The cost benefits come in two forms. Firstly the elimination of the demodulator, which has traditionally existed in discrete silicon form, and secondly the ability to deliver a global platform to OEMs, which has enormous benefits in terms of economies of scale and simplified purchasing logistics.

The basic principle that underpins the software modem approach is that compute performance is rapidly increasing for reasons that are completely unrelated to Mirics’ software demodulation. Huge computing demands are being placed upon portable and fixed devices for multimedia applications. On-the-fly video transcoding, the ability to run Adobe Flash and Java etc require significant computing power and the trend for ever more complex and demanding application software is firmly established. Improvements in silicon technology (underpinned by the so-called “Moore’s Law”), drive the reduction in power consumption at a given CPU clock speed and improvements in CPU architectures and instruction sets increase performance.

If we take DVB-T as an example, the demodulation of DVB-T on a 2.2 GHz Core2Duo CPU two years ago might have required nearly 50% of the available compute resource for this processor. Now, we see that with a 2.2 GHz Sandy Bridge, Intel’s second generation Core processor, the CPU load is comfortably under 10%. This reduction has come largely from architectural improvements in the CPU.

Intel and ARM are fighting a battle for supremacy in the discrete and embedded processor space for consumer applications such as set-top boxes, home media gateways, tablets and smartphones. The ARM CPUs had a clear advantage in terms of power consumption and the Intel x86 family currently holds the high ground in terms of performance. Both companies, along with others in this space, are investing billions of dollars each year in R&D investment and the Mirics solution benefits directly from the results of this ‘arms race’.

We have already seen this approach succeed in other areas. For example, in the 1990s, data modems were implemented in hardware and it was a very lucrative business for some. That changed rapidly when the CPU power evolved to the point where they could be implemented in software.

While, for example, demodulating a DVB-T2 signal on a 1 GHz ARM processor may not be feasible today, we would argue that it is merely a matter of time before it is and that day will be sooner than many people think. Apple’s adoption of a dual core processor in its iPad 2 has more or less guaranteed that the chasing pack of tablet makers will move to multicore CPUs in order to compete. Tablets and smartphones equipped with TI’s OMAP 5 applications processor will be able to take advantage of four ARM cores. And it is products featuring these types of next generation processor – which are being developed now – that will demonstrate the true potential of software demodulation.

TV standards evolve and become more demanding over time. DVB-T2 is substantially more complex and challenging that DVB-T. However, a new generation of TV system only comes along every 15-20 years, so consider the changes in CPU performance that can occur in that time-frame. If you need help here, 20 years ago Intel proudly rolled out its latest flagship CPU, the 486 SX, which had a top clock speed of 33MHz.

Further, in line with Moore’s Law it is clear that the integrated functionality and performance of semiconductors continues to improve exponentially, and at the same time, the prices of these devices continue to erode.

These trends allow manufacturers to continually improve the capabilities and performance of products ranging from computers and mobile phones, to set top boxes and TVs without prohibitive cost penalties.  What’s more, the performance and price points allow for the most basic of devices – from toasters to disposable pregnancy testing kits – to integrate some form of semiconductor processing.

Designing and creating semiconductors is both resource-intensive and costly, but the combination of pressure on pricing, plus the number of parties between the semiconductor design, and the end user product, makes realising the value from investment more and more difficult.  This is why companies are looking to raise their position in the semiconductor value chain, which, ultimately, necessitates getting closer to the consumer.

Of course, consumers are not going to buy chips, and in almost every case they don’t truly care or know about semiconductors. They just want to know that the end product they buy delivers or gives them access to the service or services they want with minimum or, better yet, no complications. By playing a bigger role in supporting service delivery – for instance, enabling access to mobile TV on a Smartphone – companies can add more value (and, thus contribute more to the bottom line) than they could by just providing the underlying ICs and software.

For a company to evolve to address end user needs requires an embrace of new business models with a proper re-assessment of the supply chain.  We know that end user purchasing habits are very different to the decision making units (DMUs) of traditional OEM customers.  Semiconductor companies may benefit from co-branding with OEMs and collaborating with companies in the development of enabling software, and from there it is not too much of a leap to imagine companies that started life as semiconductor designers and manufacturers supporting an ‘app’ style business model, in which the enabling software can be purchased or downloaded directly by the consumer – possibly at the same time as they sip their tall, skinny Starbucks latte.

King of the smartphone crop is of course the iPhone, which has single-handed redefined the user’s connected experience, and has set the bar [very high] for other manufacturers to reach. There is now intense competition between manufacturers to take a slice of this fastest growing cellphone sub-segment. Manufacturers and operators must offer not only desirable hardware – this is a base requirement today – but also support a large application software eco-system. While Apple, Nokia and Blackberry are well positioned in this regard with their own ‘app’ stores, Google’s Android platform is helping to level the playing field with a truly open platform, attractive to a global developer community.

The network operators who won the initial exclusive iPhone deals initially basked in their success as subscribers switched operators simply to access the iPhone, and in doing so accepted higher tariff packages; as a result, average revenue per user rose. The operators sans iPhone exclusivity sought other phone makers to compete, but most lacked the brand appeal and content eco-system of Apple. However, this situation is changing, in large part due to the open, competitive environment fostered by the Android platform.

However, a large fly in the operator’s ointment has recently become very apparent – cellular capacity crunch. The growth in smart phones, coupled with the explosion in video-based social media activity such as You Tube and Facebook, is putting immense strain on the mobile networks. A mind boggling number recently relayed to me from a presentation by Alcatel-Lucent is that 0.1% (~2000 individuals) of O2’s smartphone users consume 36% of its data capacity….and the crunch is far from over: industry commentators are predicting between 20x and 40x increase in mobile data usage over the next 4-5 years.

The operators are of course in large part responsible for this situation – by offering all-you-can-eat data packages, they have now been caught off guard by users that have decided to dine at the table of data plenty as they have become addicted to the mobile online experience. Operators have failed to develop business models that generate sufficient revenue from data usage, with a recent survey by Morgan Stanley of US operators finding that 78% of revenue was coming from voice calls, but 69% of the traffic was data. Many network operators around the world, including AT&T and O2, are now limiting downloads for new subscribers in order to contain congestion.

Can this capacity bottleneck be solved? Well, the move to next-gen standards such as LTE will no doubt help capacity (2x improvement), so to will the use of receiver diversity in handsets as already used in 3G dongles (2x improvement), and interference cancellation at cell edges will also help (~1.3x improvement). More spectrum is also key, but even here in the UK this will only realistically deliver a further 2x capacity (assuming regulators are generous with releasing ‘digital dividend’ spectrum). However, the maximum capacity increase benefit will be delivered by smaller macro cell sizes and the use of pico- and femto-cells. In this regard, a quite frightening fact is that in Japan, to deliver the ubiquitous mobile internet experience users there demanded, operators deployed 7x (seven!) the number of 3G base-stations than we have here in the UK, though the population densities are broadly similar. Maybe this vast capital expenditure explains why Vodafone retreated so unceremoniously from the Japanese market a few years ago?? In any case, this capacity issue will not solve itself; more efficient cellular technologies like LTE and diversity techniques and more spectrum will help, but they will not deliver the orders of magnitude difference in capacity that smaller cell sizes can; operators will HAVE to spend money on deploying more infrastructure, and devise user-acceptable tariffs to recoup their costs.

As operators struggle to put the data genie back into the lamp, alternative data and content delivery methods will become more important. Wi-fi on fixed networks is already popular for no/low-cost VoIP video calls and web browsing. Traditional broadcast networks offer an almost infinitely scalable mechanism for delivering live video and radio content, unencumbered by internet or cellular bandwidth constraints. Consequently, multiple content transmission mechanisms will need to coexist in portable devices to support new and existing multi-media applications to deliver the ultimate user experience while maximising efficiency of spectrum and network capacity.

However, most peoples’ portal into the WWW is via their residential gateway or mobile broadband provider and that brings me on to the subject of broadband connection speeds. I have to confess that I find the whole broadband connection speed debate a trifle irritating. Whilst there is no doubt that upgrading from a dial up modem with a few tens of kilobits per second connection speed to something that can provide many hundreds of kilobits per second or even several megabits per second brings enormous benefits in terms of data throughput, it is of course never as simple as the marketing hype from the various service providers would prefer to suggest.

In any data transmission system, there exists bottlenecks to data throughput. If all you have today for your internet connection you have a 1200 baud modem, then the chances are that the limitation lies in the modem. On the other hand, of you have a 40 megabits per second optical connection, the chances of being able to fully unlock the potential of this bandwidth is minimal. I personally have a 8 megabits per second ADSL connection at home and I am rarely if ever limited by the bandwidth of my connection at home. Indeed for most downloads, I find that the data throughput is rarely greater than a few hundred Kilobits per second. The reason for this is that my DSL connection is NOT the bottleneck in the link.

Similarly, my empirical observation is that my 7.2 Megabits per second HSDPA connection for my notebook PC, gives me much slower download times than my similar bandwidth ADSL connection. The fact is for all data that we receive, there are servers that have to deliver the data and then transport it through the network. The performance of these elements within the system are just as important to our experience as the bandwidth of our own connection. It is rather analogous to the fatuous mega-pixel race that occurred within the digital still camera market. It is in our nature to assume that ‘more’ will automatically man ‘better’, when unfortunately this may not be the case. Putting a 10 mega pixel camera into a cellphone might be a good marketing gimmick, but it when you have extremely small sensors with limited dynamic range and crude plastic lenses, the result is the same very poor quality image that you might have got from a 1 mega pixel camera.

The big content ‘driver’ for the internet is almost certain to be video. Services such as YouTube, Netflix and other video on demand services are gaining remarkable levels of popularity. However, they are also starting to create a serious problem. You might have a 40 megabits per second fibre based internet connection, but that will not mean that you will be able to automatically stream real-time HD video into your living room. The infrastructure within the wider network needs to be able to handle such throughput and the crucial question is “Who bears the cost of the infrastructure upgrade?” Where is the motivation for ISPs to bear the cost of major infrastructure upgrades just because the BBC’s iPlayer or Hulu have become so popular?

Similarly, mobile broadband providers are finding that YouTube and such as like are swallowing up vast swathes of RF spectrum and server bandwidth, but provide no commercial benefit to the connection service provider. To better illustrate the point, uniquely – among OECD countries, America has adopted no policies to require owners of cable infrastructure to open their infrastructure to competitors (c.f. BT in the UK) to enhance competition. Instead, the US relies entirely on competition based upon provision of rival infrastructure. This approach has failed wholeheartedly, demonstrated by the fact that in a recent study by Saïd Business School looking at broadband quality, Chicago, America’s best performing city, ranked 26th in the world, below Sofia and Bucharest. The FCC argues that in its recent March 16th proposal to auction large chunks of radio spectrum for mobile broadband use will provide the needed competition, but arguably this is two sides of the same coin.

Growth in demand for IP based video over the next few years is likely to require network infrastructure upgrades of several orders of magnitude.  Something is going to have to change or consumers are likely to become rather disillusioned with the reality that lies behind the marketing hype.

Hello, and welcome to the inaugural Mirics blog – MiricsPULSE – penned by me, Simon Atkinson, Mirics CEO. On this blog, I will share my thoughts on developments in the electronics industry that interest me, and I hope you will also find these musings interesting.