Hi, Lee

  In 1999, multiple infrastructure companies were transitionig from analog IF with digital basebands to digital IF and software-defined basebands.  The effects of Moore's law and the success of the first of these products in reducing OPEX have moved the markets to the point where now that everybody is using SDR infrastructure, it no longer makes sense to claim it is a proprietary trade secret to which NDAs apply.  Of course in 99, I was under NDA so I could not share this knowledge with the SDR community.  Cognitive radio has been on a similar path for heterogeneous network integration (versus dynamic spectrum), so I think we will see more announcements as software value chains reflect the integration of wired and wireline networks along with consumer hot spots (WiFi), home RF, and femtocells.  The research topics or DARPA-hard parts of integration have to do with policy languages (top-down integration) and flow interoperability (bottom-up integration).  Stevens just won a program with U. Roma la Sapienza for mathematical foundations (in semantics) and standardization of such a glue language in which policy (e.g. spectrum use), business logic (e.g. what is gold service and who gets it), and flow reconfigurability (especially for infrastructure) can be expressed interoperably.  Although the Forum's MLM group (Mitch Kokar and Rachel Li) have worked on this, their results show severe limitations of the semantic web approach.  Looking back at GSM, it was CCITT Z.100, the specificationi and description language that provided the unambiguous algebra of state machines and message sequence charts that was a pillar of GSM - $1.2Trillion in its first 10 years.  The heterogeneous systems glue langauge that we are developing with the EC should have comparable impact on heterogeneous networks.  Stevens is interested in identifying US companies and universities who might wish to participate in the US side of a joint US -European Community program to expand this Cooperation in Science and Technology (COST) action, awarded two weeks ago at $44M Euros.  The SDR Forum seems like an ideal organization for heterogeneous SDR network integration language standardization, so I thought this blog might help spread the word and to gage interest on the US side of the global marketplace.  Thanks for initiating the blog.

joe

Dr. Joseph Mitola III

Stevens Institute of Technology

Hi Joe.

I found your comments quite interesting. I can assure you that most information regarding an infrastructure vendor's basestation architecture is considered by them to be highly confidential. I know of several more companies using SDR as an enabling technology for their multi-mode basestations, but since they haven't publically announced that, I am bound by NDAs not to disclose that information.

What I have found interesting is that while the usage of the term "SDR" is noticeably increasing again (we can both recall that even a few years ago SDR had been overhyped and was a bit of a taboo word in commercial wireless), we are also seeing the rapid emergence of the terms "multi-mode", "multi-standard" and "common platform". While SDR is not necessarily required, it is more often than not the enabling technology used to make them possible.

I also find the usage of the terms quite interesting. In general, I have found that multi-mode is generally used as a marketing term to promote a differentiating capability and benefit. In fact, now it's really more table stakes. Common platform, on the other hand, I have found is generally used by engineering in reference to an internal strategy to use SDR to lower development costs while still being able to service many markets with different air interfaces.

Cheers,

Manuel

I have also found another interesting trend concerning usage of SDR by infrastructure vendors by geography. At first I couldn't quite put my finger on why this was happening, but I now have a theory which I believe people might find interesting. Many of the top Tier 1 infrastructure vendors have developed ASICs for both the radio and channel cards in their basestations previously (up to W-CDMA). However, for LTE, in general, they are foregoing ASICs due to the NRE and particularly since the cost of reconfigurable/reprogrammable processing technologies like FPGAs and DSPs have closed the gap.

But what is interesting is that some of the smaller players in areas like Japan and S. Korea (depending on your definition, they could be deemed to be Tier 1s or 2s) are still aggressively targeting ASICs and developing unique platforms for different target markets, rather than take the common platform approach. How can they justify this approach when companies with larger volumes cannot? I now believe this is primarily due to 2 factors:

1)  The requirements from the operators in those regions don't benefit as much from multi-mode BTSes as their legacy requirements are different from those in Europe, for example, where backwards compatibility all the way back to GSM is mandatory.

2)  If you look at those companies and then look at which infrastructure vendors are part of larger companies that have their own foundries, you can come to an interesting conclusion. Since those companies have foundries which need to keep running at capacity to have any chance of being profitable, I believe they are benefitting from internal pricing which is why they can still afford to do ASICs even when there are  companies doing much more volume that can no longer justify ASICs.

I'd be very interested if other folks have any thoughts or comments as to whether I'm on the right track or off in the weeks.

Cheers,

Manuel

Lee,

http://www.macom-wireless.com/products/p25/MASTR®%20V%20Base%20Station.asp

In regards to your first question, reconfigurable base stations are NOT transitory and in fact are the result of the evolutionary process of using the best technologies, based on design tradeoffs, to meet our customers' requirements.