SpectrumTalk has moved!

25th Anniversary of FCC Decision Enabling Wi-Fi and Bluetooth

25th Anniversary of FCC Decision Enabling Wi-Fi and Bluetooth
A series of posts describing how this all came about. (Click on picture above)

Thursday, December 18, 2008

Harmful Interference: The Definitional Challenge

The basic goal in spectrum management is to maximize use of the spectrum subject to limiting interference to an acceptable level. Now Marconi’s first radio didn’t have to worry about this because there were no other radios in the world to receive. However, every radio system built since then, starting with Marconi’s second radio had to worry about interference. As engineers we know that every realizable system must have a frequency response that is not perfectly band limited. Thus transmitters have some inevitable out of band emissions and receivers have inevitable susceptibility to out of band signals. In general, more expensive and more carefully designed systems have low levels of emissions and susceptibility.

Spectrum managers can arrange band plans to lessen the requirements on systems to minimize interference. But the key questions are when does interference become harmful and trading off the benefits of new wireless service with the “costs” of increased interference to previous spectrum users. (Of course, in the eyes of a long time spectrum user a new radio service that competes with his radio service has few if any benefits.)

The issue of harmful interference has been at the heart of many very controversial FCC spectrum policy decisions in the past decade. Since there is now little "green field spectrum", many new systems can only be implemented by using bands that may/might impact in some way existing users. Thus harmful interference questions have been the crux of many spectrum issues including UWB, Northpoint, PCS H Block, and AWS-3. It is fair to say that in these proceedings the opposing parties had very different views of what constituted harmful interference.

FCC, NTIA, and many other countries use the ITU Radio Regulations’ definition of “harmful interference”:
Interference which endangers the functioning of a radionavigation service or of other safety services or seriously degrades, obstructs, or repeatedly interrupts a radiocommunication service operating in accordance with [the ITU] Radio Regulations.
This definition deals with both safety-related systems and other systems. Taking out the words dealing with safety issues we get the following remaining definition of interference to other systems, such as normal commercial systems,
Interference which … seriously degrades, obstructs, or repeatedly interrupts a radiocommunication service operating in accordance with [the ITU] Radio Regulations.
But what does this actually mean? Like beauty, it is often in the eye of the beholder. When the ITU adopted this definition decades ago, the upper limit of usable spectrum was much lower than it is today and propagation was generally either ionospheric or involved relatively simple tropospheric concepts. However, we now use frequencies subject to many additional propagation mechanisms such a rain fading, multipath fading, and atmospheric absorption. Some of these cause outages independent of interference. So is it appropriate to focus on marginal interference due to other radio systems without some reference to naturally occurring outages? Frankly, there is no consensus on this issue.

The FCC’s 2002 Spectrum Policy Task Force (SPTF) examined the issue of defining interference, holding public fora and accepting written comments. SPTF observed,
Several parties also pointed out that the Commission’s most difficult, controversial, and unsatisfactorily resolved cases have resulted from situations in which the extent of an incumbent’s spectrum rights and interference rights, and its limitation on impacting other bands or users, were not clearly understood by the incumbent, by a new service provider, and even by this Commission.
Thus uncertainties about interference protection affect all parties involved and may inhibit development of new radio systems and investment in existing systems. Improved certainty, what ever it is, would benefit all parties.

FCC explored this direction in the ill fated “interference temperature” proceeding. It boldly stated,

This new concept could shift the current method for assessing interference which is based on transmitter operations, to an approach that is based on the actual radiofrequency (RF) environment, taking into account the interactions between transmitters and receivers.

The interference temperature model could represent a fundamental paradigm shift in the Commission’s approach to spectrum management by specifying a potentially more accurate measure of interference that takes into account the cumulative effects of all undesired RF energy, i.e., energy that may result in interference from both transmitters and noise sources, that is present at a receiver at any instant of time. This new approach could provide radio service licensees with greater certainty regarding the maximum permissible interference, and greater protections against harmful interference that could be present in the frequency bands in which they operate.
However, FCC confused this goal by also proposing in the same document additional unlicensed use as underlays in most bands to take advantage of situation where licensees were receiving “too little” interference, bringing them up to the proposed guaranteed level. Clearly, no incumbents thought they were receiving too little interference and saw this combined approach as a threat. With a crescendo of criticism and no strong supporters the proposals – both the defining of interference and the proposed additional unlicensed use - were killed off in May 2007.

Fortunately, the issue of clarifying what is interference is continues on the other side of the Atlantic. On October 9, 2007, the European Commission held a workshop in Brussels on interference management. A key part of the workshop was discussing a study by Eurostrategies sprl and LS Telcom that was sponsored by the European Commission. The study examined whether spectrum could be used more effectively by controlling interference itself, rather than relying on the traditional method for achieving this: limiting the technical parameters of transmitters. The study is not yet completed, but is indicative of European attempts to consider interference as a basic issue.

As demand for wireless services increases, there is little “green field” spectrum available. Increased spectrum utilization can be achieved by more efficient technologies by existing users in many cases. But in other cases novel spectrum sharing may be necessary to increase spectrum utilization. Recent growing interest in time division duplex to implement asymmetric data flows in mobile networks also raises novel interference questions. The ambiguities in present understanding of the interference protection rights affects both incumbents and innovators. Hopefully more discussion within the technical community about how to quantify interference expectations will be helpful to all involved.

UPDATE
1/30/09
Mitchell Lazarus has a blog post on this topic also.

He ends with he following conclusion:

"Given the wide range of valid considerations in play, no fixed definition of “harmful interference” can give the right answer in every instance. Any such definition would run a very high risk either of letting in new services that in fact are likely to disrupt existing applications, or of unnecessarily locking out useful technologies that could be accommodated without harm.

Besides, innovation is hard to predict. (If predictable, it wouldn’t be innovation.) Even if we could somehow formulate a definition that appropriately covers all of today’s cases, chances are that some new technology would soon raise questions that the definition cannot properly answer.

The present definition, vague though it is, allows the FCC to reach the right result in difficult cases. And nowadays, most of the cases are difficult."

While I don't see a perfect ultimate definition, I think the current definition is a real transparency issue. It can be improved.



2 comments:

oded418 said...

It's time for SPTF-II, especially with a new leadership FCC. Personally, I don't think that quantified, measurable and enforceable "harmful interference" is possible without taking into account the heretofore ignored receiver topology.
In other words The FCC could give more flexibility to manufacturers that produce Rx less prone to interference provided the Rx meet certain minimum specs.

Michael Whittaker said...

The Australian Regulatory Model (Space-centric Management)

The cited December 2007 report on radio interference regulatory models was released by the EC in April 2008. It explores “the possibility of using interference definitions as a method of defining the rights of spectrum users in a liberalised environment.” The report “suggests that a single universal definition of harmful interference suitable for all applications and technologies is unlikely to be realisable”. This result was not unexpected given the many years the ITU wrestled with trying to quantify and give practical meaning to ‘harmful interference’. The only situation where the ITU had a measure of success was with the simple case of the satellite service.

For flexible spectrum access, an interference level which causes problems for one technology may be inconsequential for another. Therefore, spectrum rights based on unacceptable levels of field strength/aggregate power flux density are not practical. Interference temperature turned out to be more a cause for embarrassment for the USA Task Force than a solution. Similarly, Ofcom’s attempt to “specify in a licence the interference a licensee is allowed to cause - SURs”, began with an already long unsuccessful history and has been rejected by most of UK industry. Both the USA and Ofcom proposals can not produce practical outcomes for spectrum licensees.

What a legal right comprises depends on what is said by what confers it. A right may be conferred ‘positively’ or ‘negatively’. Lawyers have for some time recognised it is much more practical in drafting terms to establish the content of a right by defining it negatively i.e. permission is conferred to use the spectrum subject to certain restrictions, rather than trying to describe the extent of the right in positive terms. Therefore, explicit (primary) transmit rights with implicit (secondary) receive protection is more practical.

In principle there is a logical nexus between transmitting and receiving with one element not being more important that the other, but in practice, limits at a transmitter(s) antenna or antenna array, which have been designed to embed a predictable level of receiver protection in relation to all interference mechanisms, are clearly superior because they lead to much better managerial efficiencies.

In early 2008, the EC adoption of a transmit based limit (BEM) for introducing spectrum liberalisation throughout Europe (WAPECS) provided further supporting evidence as to the superiority of explicit transmit rights. BEM is now rightly preferred by European industry over Ofcom’s proposals for primary limits of aggregate power flux density.

The BEM of WAPECS was a good start for increasing equipment choice in Europe. Unfortunately, BEM deals only with one aspect of interference at the frequency boundary. It is a partial solution. Similar benchmarks are needed for all the other interference mechanisms.

The provision of partial spectrum rights and/or rights based on explicit receive parameters is the primary reason why there is so much regulatory uncertainty in both Europe and the USA, uncertainty that has never been experienced under the Australian regulatory model.

Australia employs a complete set of transmit rights. For new services, the transmit rights are established as mathematical functions of device separation from boundaries of a spectrum space i.e. geographic, frequency and time. For legacy services, separate device-specific and site-specific receive rights which can override the explicit transmit rights legacy services are usually used for political reasons.

A complete set of explicit transmit rights in relation to all interference mechanisms enables a licensee to authorise and operate devices completely independently of the regulator and adjacent spectrum licensees, and if desired, without a formal equipment standardisation process. With a complete set of rights, any type of new equipment can be independently authorised by a spectrum licensee essentially in the time it takes to make a minimum number of laboratory measurements and check its field deployment against the spectrum access conditions of the licence.

A high level of engineering skill is necessary to establish a coherent set of benchmarks to achieve self-consistent levels of notional receiver protection. Australia provided such benchmarks in 1997 and they have been successfully used in practice to introduce innovative wireless services into Australia for over a decade, without any reported cases of interference or related litigation.

Further information can be found at http://www.futurepace.com.au/_lib/pdf/DSA.pdf