Breaking the 5G Curve by Looking Beyond the U.S. Patent System

By Shawn Ambwani & Sam Jaffna
December 16, 2020

“Manufacturers of equipment implementing 5G technologies should consider strategic oppositions and invalidity proceedings outside the United States against SEPs held by major licensees.”

A wave of thousands of 5G Self-Declared Standard Essential Patents (SD-SEPs) applicable to everything from devices to network infrastructure is fast approaching. The graph below shows the viral growth in the number of 5G self-declared essential applications published on post-2000 inventions.

Chart produced from Unified Patents Portal

The value of these patents is 6-10% of the retail product value, if recent LTE SEPs court decisions are to be believed. However, ex ante 5G licensing rates announced by traditional licensors Qualcomm, Ericsson, Nokia, and Interdigital total around $18 (or 3.6%) on a $500 handset. Yet these licensors hold less than 17.4% of the relevant 5G SD-SEP families, which would make the total royalty burden 20% or higher.

Implementers faced with high SEP licensing cost and uncertainty typically mitigate risk by: (1) using licensed components, (2) receiving indemnification, and (3) leveraging defensive portfolios. But there is another strategy that should be considered given the tools which are now available: preemptively challenging patent family validity in foreign jurisdictions that are relatively quick, inexpensive and often more effective.

SEP Tidal Wave

There are 94,000 active grants and applications (25,700 families) 5G SD-SEPs as of September 2020. Of these, almost two-thirds (64%) have not yet been granted. If LTE is indicative of the future for 5G, the numbers will continue to compound. LTE SD-SEPs grew almost 5% annually from 2011 to 2020. Assuming growth for 5G SD-SEPs is the same, this means by 2030 there would be more than 150,000.

Chart produced from Unified Patents Portal

Calculating the 5G Device Landscape

Not all 5G patents apply to devices (in 3rd Generation Partnership Project parlance called “User Equipment” or “UE”). Using CPC codes to patents litigated against LTE devices and sampling for effectiveness, we identified about 79,800 5G patents and applications (12,300 families) that could apply to UE.

As with LTE, we can expect many 5G SD-SEPs will be found non-essential when challenged. Not surprisingly, there has been very little consensus on this. In the Unwired Planet v. Huawei (UK) LTE trial, experts differed on essentiality by almost 22% and the court held about 25% of SD-SEPs were essential. Costly eyes-on SEP reviews continue to be criticized for being: (i) too keyword reliant, (ii) having substantial sampling bias, (iii) using unqualified experts and/or not providing enough time, (iv) not taking into account undeclared patents, or (v) not weighing patents differently based on their perceived value to the standard.

Unified Patents’ AI-based Objective Patent Landscaping tool (OPAL) uses all 5G SD-SEPs as a positive and all 2G, 3G, or LTE SD-SEPs not declared in 5G as a negative training set. A higher OPAL score means a publication is more semantically similar to the training set.

Chart produced from Unified Patents Portal

Filtering for semantic similarity above 0.75, OPAL identifies around 9,700 5G SD-SEP UE families that most likely to be 5G. In addition, OPAL also identified 5,000 other families not self-declared at the same similarity.

Top 20 Holders of 5G Self-Declared UE Families in US at >Similarity 75

Chart produced from Unified Patents Portal

OPAL identified about 5,000 5G UE SEP families with a U.S. counterpart and of those about 75% have a European Patent (EP), Japanese Patent (JP), Korean Patent (KP), or Chinese (CN) counterpart. Non-U.S. counterparts represent an invalidation risk to their U.S. family members due to quicker and more effective foreign validity challenges. A successful foreign opposition provides strong arguments for showing invalidity in the United States and other countries.

Top 20 Avanci Members per their 5G SD-SEPs UE Families in US at >Similarity 75

Opposition and Nullity Proceedings Outside the United States

Most countries offer post-grant opposition proceedings to improve patent quality. In the United States, the inter partes review (IPR) process was established under the 2011 Leahy-Smith America Invents Act (AIA). An IPR’s high costs ($300,000 to $500,000) have prevented invalidity challenges in that forum from becoming as prevalent as European Patent Office (EPO) oppositions. In fact, 6% of all EP grants are opposed, while only 1% of all U.S. grants are challenged.

The EPO, Japan and Korea today offer effective opposition processes. Challenges outside the United States are often more cost effective than U.S. IPRs. In some cases, 6-10 patents can be challenged outside the United States for the cost of a single U.S. IPR. China does not have an opposition process but its invalidation process is quick and open, serves the same result and provides more flexibility. The table below compares foreign country opposition procedures.

Effects of Successful Foreign Oppositions

The value of major 5G portfolios can be affected by 5G challenges in the EU, Japan, Korea and China.

As a case in point, Avanci’s LTE members, likely candidates for a 5G pool, hold around 5,200 5G SD-SEP UE families. About 50% of those have no US publication which usually means a lower licensing valuation. Almost 80% (1500 families) of the 1,900 could be challenged in the EU, Japan, Korea, and China.

While successful validity challenges of a patent in the EU, Japan, Korea and China do not conclusively prove the invalidity of the U.S. counterpart, the successful arguments used in these challenges have been effective in the US and used to argue lower portfolio values.

Studies show challenges also reduces patenting by 15% by the same or similar entities. In a 2019 study conducted by Markus Nagler and Stefan Sorg on the basis of 30,000 EP oppositions from 1994 until 2010, showed successful ones reduced the volume of future filings by the same inventor.

Look Overseas for Success

Manufacturers of equipment implementing 5G technologies should consider strategic oppositions and invalidity proceedings outside the United States against SEPs held by major licensees. These challenges cost a tenth or more less than typical U.S. challenges and could impact the licensing value of up to 50% of the self-declared families of the major SEP holders. Successful challenges also could produce the added future benefit of reduced patenting and increased patent quality.

Image Source: Deposit photos
Image ID:182552442
Copyright:jamesteohart 

The Author

Shawn Ambwani

Shawn Ambwani is COO / SVP Legal / Co-Founder at Unified Patents.

Shawn Ambwani

Sam Jaffna is an attorney and a general manager at Unified Patents. He has been working at Unified Patents since 2014. He received degrees in math and physics in 2007 from Rice University and received his Juris Doctorate in 2013 from Harvard Law School. Prior to joining Unified Patents, Sam worked as a software developer at a financial company and then as an engineer at a satellite company.

Warning & Disclaimer: The pages, articles and comments on IPWatchdog.com do not constitute legal advice, nor do they create any attorney-client relationship. The articles published express the personal opinion and views of the author as of the time of publication and should not be attributed to the author’s employer, clients or the sponsors of IPWatchdog.com. Read more.

Discuss this

There are currently 12 Comments comments. Join the discussion.

  1. Anon December 17, 2020 7:24 am

    Can you delve into this “self-declared” notion a bit?

  2. Skeptical December 17, 2020 8:43 am

    Despite Gene’s argument that math is not science, I found myself doing quite a bit of math in my undergrad eng physics courses (firmly and ironically in Cat. A, I might add). It makes me uncomfortable to see so many statistics with unexplained confounding variables.

    But the one I most don’t understand here is why semantic similarity has anything to do with the “quality” of a patent. Is there data to support the notion that semantically similar patents are “bad?” It seems like it could depend on many arbitrary factors, such as the innovation focus/breadth of a given company and how they draft their specs. And aren’t their differences in languages that could make that metric prejudicial to certain countries/languages?

  3. Shawn Ambwani December 17, 2020 2:03 pm

    Good points. Semantic similarity is not an indicator of validity. We do have a tool called APIX which provide some measure for it. It is an indicator based on the word and phrases of the training set which indicates essentiality. You are right that there is an assumption that many of the patents are not valid, but that is just based on historical norms of validity when challenged.

  4. Anon December 17, 2020 2:24 pm

    There is an assumption… based on…

    Sound like “we see what we want to see” (never mind the legal presence of the statutory presumption to the contrary of the assumption built into the models being used…

    One reason why I asked the question at post one is that there appears to be a lot of assumptions built in.

    Would be nice to detail those out (of for no other reason than to be able to develop a sense of robustness by plugging in ‘counter-assumptions’).

  5. __tr__ December 17, 2020 3:31 pm

    @Anon, the concept of self-declaration just means that the SSO is not normally what defines what patents are essential; the assignee makes a declaration. For instance, according to ETSI’s IPR Policy, Rule 4.1:

    “a MEMBER submitting a technical proposal for a STANDARD or TECHNICAL SPECIFICATION shall, on a bona fide basis, draw the attention of ETSI to any of that MEMBER’s IPR which might be ESSENTIAL if that proposal is adopted.

    Because the scope of a given standard will change by the time actual implementation, an SSO or assignee can’t evaluate with certainty whether a patent will end up being “essential” to a given standard. There is unsurprisingly a tendency to over-declare as it ensures that a patent will be subject to a royalty stream.

  6. Anon December 17, 2020 5:35 pm

    Thanks __tr__ – in other words, ‘self-declaration’ has NO real actual effect, and is merely a mechanism by which something MAY make it to the status of SEP.

  7. __tr__ December 17, 2020 6:46 pm

    @Anon, the whole process turns into somewhat of a pile of dust from there and sometimes leads to head-scratching results. I’m not going to pretend to be an expert on the topic because I’m not, but my understanding is that the SSO can choose whether or not to accept a patent to a standard, and if so, the patentee must make a FRAND commitment. But that does not mean that the patent is necessarily part of the standard. Likewise, rejection by an SSO does not mean that a patent is not essential to a standard (however, in that situation, if a patentee goes on to accuse an implementer of infringement, they are opening themselves up as a rogue SEP and will likely be challenged).

    The FRAND commitment itself is a hairy topic because the SEP holder does not need to disclose their royalty rates before being accepted by most SSOs and can arguably still hold up if an implementer is unwilling to give them a “FRAND rate” (which is still not a concrete concept, no court has really stated what a FRAND rate is). However, the SD-SEP holder could also join up with a patent pool that purports to implement some portion of an SSO’s standard. Once in a patent pool, essentiality may not matter because the implementer pays a bundled rate for a license to the pool. As one might guess, the idea that patents in a pool may not actually be essential to a standard, and therefore infringed by an implementer, creates a host of antitrust problems that are still shaking out (although the FTC has released fairly extensive guidelines that are considered to be more or less “safe harbors” for how patent pools can legally operate).

  8. Anon December 17, 2020 8:35 pm

    Thanks __tr__, that was indeed helpful.

  9. Shawn Ambwani December 18, 2020 12:19 am

    TR. All good points. The methodology for all the tools mentioned are on the site. There is no perfect tool at this point, especially when discussing SD-SEP patents since they are often never evaluated independently for essentiality. This creates a huge problem since the alternative (which no one has been willing to do) is eyes on evaluation of all patents. So we created the next best thing. Use the SD-SEP and a negative training set to create a similarity measurement to the semantics of the patents most likely to be essential. It is not determinative, but better the alternative in untested areas such as 5G.

  10. __tr__ December 18, 2020 8:30 am

    @Shawn Ambwani, thanks for the explanation. Now I understand what’s going on here. Yeah, I mean, there’s probably some merit to the semantic similarity in eliminating outlier declarants who are obviously not in the standard.

    I don’t know if evaluating the entire publication is the optimal approach because each patent likely has some amount of boilerplate language shared across 5G publications. Semantic similarity would seem to reward declarants that use more boilerplate in their spec. The training set’s size seems large enough to invoke a more sophisticated model for weak learners, such as gradient boosting against the actual claims. Such an approach would also discover features for further analysis.

    Is the database you are using pulled directly from ETSI or some other publicly available resource?

  11. Curious December 18, 2020 5:34 pm

    Manufacturers of equipment implementing 5G technologies should consider strategic oppositions and invalidity proceedings outside the United States against SEPs held by major licensees.
    There are those companies that create intellectual property, and there are those companies that destroy intellectual property. There is no secret as to where Unified Patents falls.

    As most parents of children can attest to, it takes small children anywhere between the ages of 5-7 before they grow out of the phase where they want to destroy what other people build. It is must easier to destroy than to build at that age, and destroying things gives these small children a sense of power that they wouldn’t ordinarily have.

    It makes for an interesting comparison, don’t you think?

  12. shawn ambwani December 18, 2020 5:47 pm

    Your points are again well taken. We pull the SD-SEPs from ETSI. There are ways to improve the scores based on a number of different vectors. The big question is how do we know if it is better.

Post a Comment

Respectfully add to the discussion.

Name *
Email *
Website