Unwired Planet International v Huawei Technologies Co.

Mr Justice Birss gave an important decision on 5 April 2017 concerning FRAND undertakings. FRAND stands for ‘fair, reasonable and non-discriminatory’ and the FRAND undertaking is the undertaking a patentee has to give when declaring a patent to be essential (‘essential’ meaning that it would inevitably be infringed by operating in accordance with a given standard) to the standards setting organisation (SSO) that it will grant licences on FRAND terms.  Until this decision it was not clear exactly what ‘fair, reasonable and non-discriminatory’ meant nor the exact legal standing of the FRAND undertaking and while this decision cannot provide all the answers for all situations, it has shed some light on how the FRAND undertaking operates.

This action started in April 2014 when Unwired Planet sued Huawei together with Google and Samsung for infringement of six patents, five of which were claimed to be essential.  The cases with Google and Samsung settled but Huawei and Unwired Planet failed to come to an agreement.  Rather than try to deal with all issues in a single trial, the dispute was split into its issues with a technical trial to determine the validity, infringement and essentiality of each of the patents followed by a non-technical trial to determine the FRAND and competition issues, injunctive relief and damages for past infringements.  Only three of the technical trials were held, with two of the patents held valid and infringed, and this judgment concerns the non-technical trial.

We have previously reported on the technical trials and copies of those articles are available here and here.

Legal Effect of a FRAND undertaking

A FRAND undertaking to ETSI (European Telecommunications Standards Institute) is governed by French law, because ETSI is based in France, and the judge found, having heard evidence from French law experts, that an undertaking of this type (that is to say, from a company to a Standards Setting Organisation) is enforceable by third parties.

The same judge had previously held that if both sides are willing to be bound by the outcome then the English court could grant a declaration that the terms of a given licence are or are not FRAND (Vringo v ZTE [2013] EWHC 1591).  In this case he went a step further and held that the court could also decide whether certain terms need to be adjusted in order to make a given set of terms FRAND.  The court’s jurisdiction is therefore not restricted to the binary question of assessing a given set of terms but extends to deciding between rival proposals and coming to a conclusion different from either side’s case on such a proposal.

Once the court has decided what the FRAND terms are for a particular set of circumstances then an implementer (i.e. a company which makes hardware which incorporates Standard Essential Patents (SEPs)) which refuses to take a licence on those FRAND terms has effectively chosen to have no licence, and so if it has been found to infringe a valid patent then an injunction can be granted against it.  However, an implementer who makes an unqualified commitment to take a licence on FRAND terms cannot be the subject of a final injunction to restrain patent infringement.

FRAND as a benchmark rate

The judge found that there are two approaches to determine the FRAND licence rate: the ‘top down’ approach and the ‘comparable licences’ approach.

The ‘top down’ approach is to start with a figure representing what the appropriate total aggregate royalty burden should be for a given standard (T).  T can then be shared out across all the licensors for that standard in proportion to the value of each licensor’s patent portfolio, with the value of a share of a portfolio of the total being ‘S’.  The FRAND rate for each portfolio is the product of TxS.

The exercise, in summary, comprises identifying how many of the portfolio of patents being licensed are ‘relevant SEPs’ (i.e. relevant to the standard) and then to ascertain what percentage of the total number of relevant SEPs for the standard the portfolio being licensed represents and therefore what percentage of the total royalty burden should be allocated to it.

Identifying the total number of ‘relevant SEPs’ for a standard is more complicated than simply adding up the patents declared to be essential. The process must take into account the different jurisdictions, divisionals and the fact that many more patents are declared to be an SEP than in fact are ‘essential’. The parties in this case each used different methods, Unwired Planet used a method called MNPA and Huawei used one called HPA.  The judge decided that both methods were wrong but that using the simpler and more transparent HPA method and adjusting the results was the right way to reach a conclusion.  The HPA method consisted of the following steps:

(1)   “Identification and De-duplication”: a list of declared essential patents and patent applications was created using the ETSI database and also making reference to the Korean Telecommunications Technology Association database.  The list was de-duplicated.

(2)   “Family members not expressly declared to ETSI”: Since the ETSI IPR Policy declaration applies to a patent family as a whole, additional family members not expressly declared to ETSI were identified.  This was done using the public INPADOC database. 

(3)   “Grouping families in five categories”: the patents and applications were collected into families.  The families were collected into five groups.  Only group 1 was selected for further analysis.  The five groups were:

Group 1 – at least one issued and non-expired patent and an English or Chinese language member;

Group 2 – at least one issued and non-expired patent but no English or Chinese language member;

Group 3 – only expired members

Group 4 – no issued patents (“issued” means granted)

Group 5 – family information not available on INPADOC

(4)   “Grouping families into standards”: the families were classified into three classes: LTE/4G, UMTS/3G, GSM/2G by reference to the standards to which they were declared on the ETSI website.  The families were also classified as relevant either to RAN (which in this study includes handsets) or core network (“CN”).  This was also based on the standards to which they were declared.

(5)   “Essentiality analysis of Group 1 families”: The Evaluators reviewed the essentiality of a patent in each Group 1 family.  The review took about 30 minutes per family.  The patent and relevant standard were selected in accordance with given rules.  The claims of the patent were compared to the relevant standard specification to determine if the standard required all the elements of the claims.  If the Evaluator determines that the specification does not provide a clear reason to rule out the patent as being essential, then the family is deemed essential.  If the family provides a clear reason to rule out the patent being essential, the family is deemed not essential.  The given rules are:

a.       Patents in the family are reviewed in the following order until a patent is deemed essential or the categories are exhausted.  If multiple patents are in the categories then the earliest is looked at first.  The categories are:

i.      US issued patent

ii.     EP issued patent

iii.    Any other English language issued patent

iv.    Chinese issued patent

v.     English-language expired patent or subsequently English language application (where there is no English language or Chinese language issued non-expired member but there are members from other jurisdictions that are issued and not expired). 

b.      For each family both representative handset and infrastructure claims are identified.

c.       If the family is declared to more than one of LTE/4G, UMTS/3G, and GSM/2G then the family analysis is continued until a patent or application is found essential to each of these three standards or the categories are exhausted.”

The HPA method produced a number of 1812 Relevant SEPs for 4G handsets, which the judge felt was too many.  The Revised MNPA method produced a figure of 355 which the judge felt was too low.  He therefore held that 800 (mid-way between half of 1812 and double 355) was the correct number.  A point to note is that all valid patents in a given category are treated as of equal value.

There was an added layer of complexity in this case in that Unwired Planet had acquired its patent portfolio from Ericsson and therefore Unwired Planet’s share of the relevant patents for the particular standard was a sub-set of the overall share belonging to Ericsson before it sold the portfolio to Unwired Planet.  This affected the calculation of the percentage to be allocated.

The ‘comparable licences’ approach is also not without its complications because not all licences are usefully comparable.  Of the 11 licences proposed to be comparable, the judge was prepared to place weight on six in order to determine what an appropriate licence might be.  The judge considered all the material and decided on a representative royalty rate for Ericsson’s portfolio, and then applied a ratio representing the respective strength of the Ericsson and Unwired Planet portfolios to arrive at a FRAND figure for Unwired Planet’s portfolio.  He then cross-checked the result of that calculation (i.e. the licence rate) by calculating what the total royalty burden would be if all relevant SEPs were licensed at a proportional rate.

FRAND is non-discriminatory

The judge concluded that ‘non-discriminatory’ in the context of a FRAND undertaking means that a benchmark FRAND rate should be derived which is applicable to all licensees seeking the same kind of licence.  The benchmark rate approach to assessing a royalty is itself non-discriminatory because it is based on the strength of the portfolios, not on factors associated with the licensee.

The judge added that if, contrary to that finding, the FRAND undertaking also includes a specific non-discrimination obligation whereby a licensee has the right to demand the very same rate as has been granted to another licensee which is lower than the benchmark rate, then that obligation only applies if the difference would distort competition between the two licensees.  This was important because Samsung had in fact previously been granted a licence at a lower rate.

FRAND can be worldwide

A question before the court was whether it would be anti-competitive for a FRAND licence to be worldwide, because a worldwide licence would tie or bundle all the patents around the world into a single product sold by an entity dominant in the market.  Case T-201/04 Microsoft Corp. sets out a four point test to determine whether bundling is contrary to competition law.

Before addressing the Microsoft test, the judge asked what sort of licence a willing licensor and a willing licensee acting reasonably would agree when contemplating the SEPs the subject of the litigation.  He held that there is only one answer and that is that they would agree on a worldwide licence.  The licence might employ different rates for different countries but the licence itself would be worldwide.  However, if that worldwide licence would result in unlawful bundling then it would not be FRAND and therefore the judge applied the four point test set out in Microsoft.

The four points are:

“- first, the tying and tied products are two separate products;

- second, the undertaking concerned is dominant in the market for the tying product;

- third, the undertaking concerned does not give customers a choice to obtain the tying product without the tied product; and

- fourth, the practice in question forecloses competition.”

Only the fourth point required any detailed analysis.  The judge, in order to be able to hold that there has been abuse of a dominant position, must find that the practice in question forecloses competition.  Such a finding may be based on inference but the inference must be justified in all circumstances. In the present case, the judge drew, from the prevalence in the industry of worldwide licences and the prevalence of assessing rates based on patent families, the inference that multi-jurisdictional portfolio licences themselves are unlikely to have inherently anti-competitive effects and that a demand for a worldwide licence is not inherently likely to distort competition.  Starting from that inference, the judge analysed the actual facts before him and found that a worldwide licence would not be contrary to competition law on the facts of this case.

The terms of the licence should provide for events such as revocations or declarations of non-essentiality in any individual countries to avoid the problem of a licensee having to pay royalties in territories where there is no valid, infringed patent, and to ensure that only the courts of the country in which the patent is held have jurisdiction over validity (in compliance with Art 24(4) of the Brussels I Regulation), but the licence should be worldwide.  The consequence of the FRAND terms being a worldwide licence is that Unwired Planet was entitled to insist on that geographic scope (Huawei had argued for UK only) and if Huawei refuses to accept the FRAND licence then the English court can grant an injunction to restrain its patent infringement in the UK.


This decision cannot provide all the answers for all situations but we can take from it that:

  • a FRAND undertaking from a company to ETSI can be relied on by a third party and enforced by an English court;
  • there is only one set of FRAND terms for a given set of circumstances and if the implementer refuses to accept those terms then the court can grant an injunction against the sale of infringing products;
  • a benchmark royalty rate governed by the value of the patentee’s portfolio will generally be non-discriminatory;
  • a FRAND rate need not match a lower royalty rate agreed with a third party unless the difference is large enough to distort competition;
  • a FRAND licence can be worldwide, depending on the norm in the industry, as long as the terms are such that the licence does not foreclose competition.

A copy of the full judgment is available at http://www.bailii.org/ew/cases/EWHC/Patents/2017/711.html


12 May 2017

Search by Name or Specialism