This paper will mainly take the SEP intensity analysis in the field of communication as an example to simplify the statistical method and necessity judgment method of the number of SEPs, in order to provide reference for relevant practitioners and parties to disputes.Author |Wu Ziqing, Jia Hongpeng, Beijing Liu Shen Law Firm.
Edit |Bruce.
Standard essential patents (SEPs) refer to patents that must be used to implement specific technical standards (e.g., international standards, national standards, or industry standards) developed by standardization organizations (such as the ISO ISO ITU IEC, the IEEE, the regional standardization organizations ETSI and CEN Cenelec, and the national standardization organizations ANSI and BSI[1]). The requirements for members to disclose information and license their patented technologies in the process of joining the organization or participating in the development of standards vary from one standardization organization to another, but most standardization organizations require members to disclose their patent information in a timely manner and require licensing based on the principle of FRAND RAND. However, standardization organizations do not usually examine the necessity of these patents (i.e., their correspondence with standards).
With the promotion and implementation of standardized technologies around the world, the number of SEPs that meet certain standards (or are declared to meet the standards) has also become very large. For example, in the mobile communications industry, as of June 2022, more than 210,000 5G SEPs have been declared worldwide[2]. In this way, in the process of standard implementation, determining how many SEPs and SEP families a certain SEP right holder owns, and which patents belong to the ESP that truly meet the standards, often becomes the focus of dispute between the SEP right holder and the standard implementer when negotiating FRAND RAND licensing, so that both parties have to resort to arbitration institutions or courts to resolve the dispute. This paper will mainly take the SEP intensity analysis in the field of communication as an example to simplify the statistical method and necessity judgment method of the number of SEPs, in order to provide reference for relevant practitioners and parties to disputes.
1.SEP strength analysis in the field of communications
1.1 Based on analysis of ETSI's special report
The formulation of communication standards is generally the responsibility of international organizations such as the International Organization for Standardization (ISO) and the International Telecommunication Union (ITU) or national standardization organizations. As the world's largest and most important international communication standards organization, 3GPP (Third Generation Partnership Program) has united seven telecommunication standards development organizations (ARIB, ATIS, CCSA, ETSI, TSDSI, TTA, TTC) to develop various technical standards and specifications for mobile communication systems to ensure that devices and systems from different manufacturers can be compatible and interoperable. As the main organization of 3GPP, ETSI is responsible for coordinating the activities of 3GPP, and assumes the responsibility of the secretariat, providing the management and organizational framework of the 3GPP protocol, and carrying out daily 3GPP maintenance work.
In line with ETSI's IP Policy, ETSI requires its members to "promptly notify the Director General if ETSI members become aware of any patents that may be necessary during the proposal or development of standards"[3]. The key here is "potentially necessary patents" rather than "truly necessary patents". In addition, ETSI does not examine the necessity of these patents. This has the consequence that each member is trying to declare to ETSI as SEPs what it considers to be "potentially necessary" patents, thus creating a de facto phenomenon of "over-declaration" and the possibility that some patents that are not genuinely necessary may be declared as essential.
The phenomenon of "over-declaration" can be said to be unavoidable for the following main reasons: 1) Members generally declare early, so a large number of patent applications are declared, but these patent applications may be narrowed or modified in the course of examination to no longer conform to the standard, or may be rejected and cannot become SEP;2) With the evolution of the standard, what was originally an SEP will no longer become an SEP due to the revision of the standard.
In addition to the phenomenon of "excessive declaration", there is also the phenomenon of "duplicate declaration", which includes the following main reasons: 1) the same patent is declared successively by different patentees due to the change of patentee;2) Different patents in the same patent family are declared separately;3) The same patent is re-declared due to a change in the content of its own claims or a change in the standards involved.
Due to the existence of excessive declarations and duplicate declarations, it is inevitable that SEP rights holders and SEP implementers must analyze the amount of SEP owned by SEP patentees and their necessity when negotiating FRAND licenses.
Considering that ETSI maintains all declared patents and provides an intellectual property database that allows the public to access patents that have been declared to be essential or likely to be essential to ETSI and the 3GPP standards, the public can make use of the original declaration documents from this database or from the Special Report that ETSI has extended with the European Patent Office's patent database "espacenet". The strength of SEP patents is analyzed for the patent information contained in the declaration document or special report.
The following is an example of the process of performing an SEP strength analysis using a special report maintained by ETSI for a specific date. The analysis process is based only on the specific special report of the date as an example. The data and analysis results in this example are not applicable to other areas, but the methodology used is somewhat universal.
Step 1: Data collection
The ETSI's special report itself is aimed at providing information on the relationship between technologies and standards, and as such provides a statistical overview of the relationships between the elements included in all IPR declarations, as well as aggregated details of the declarations. The author ** ETSI's latest updated special report (updated on March 13, 2023), which includes a total of 4,244,359 pieces of patent data, of which each patent data entry includes information such as patent publication number, patentee, patent family ID (used to identify the family), declaration date, declared standard, project team, etc.
Step 2: Data cleansing and standardization
For the reasons of the aforesaid duplicate declarations, there are patents and their families that are clearly duplicate in the Special Report. In addition, there may be cases where the declaration is incomplete, making it impossible to determine the exact patent number of the claim entry, expand its family, or identify its patentee in the ESTI Special Report. Furthermore, in order to make the analysis more readable, it is necessary to standardize the claim data: for example, to normalize the related claim subjects that belong to the same group but apply for patents under different applicant names to a single name.
To do this, the first step in data processing is to clean and standardize the data in the report. For example, the exclusion of claims that have missing or incorrect patent information that makes it impossible to identify their patent information. For example, for the same patent, only one piece of patent data can be retained on the basis of merging the standards and project groups involved in multiple declarations according to the declaration date, and the remaining duplicate entries can be removed.
The author carried out data cleaning on the above-mentioned 4,244,359 patent data, eliminated the entries with missing patent information, and screened out the patents declared as 5G on the above-mentioned specific date, which can obtain 1,244,630 patent data, and further deduplicated the same patent entries to obtain 209,688 patent data, which correspond to 209,688 patents.
Step 3: Patent family
In the field of telecommunications, the general practice of FRAND licensing is to license globally or in multiple countries based on a portfolio, so it is necessary to count the patent families owned by SEP rights holders and judge the necessity of members within the patent family.
Under ETSI's Intellectual Property Policy, all current or future members of the patent family are deemed to have been declared if a member of the patent family is declared in a timely manner, but ETSI still allows members to voluntarily provide information about other members of the patent family. Here, a "patent family" (or simply a "patent family") is defined as a document with at least one public priority, including the priority document itself. The "documents" here include invention patents, utility model patents and their applications. As a result, in the ETSI Special Report, it expanded the family of the declared patents and assigned a family ID to all the declared patents.
For this purpose, all patent entries can be grouped based on the family numbers in the special report, and 53,670 5G families can be obtained from the above 209,688 5G patent data.
Step 4: Inquire and confirm the legal status of the patent family
The "legal status" of a patent can be summarized into three statuses: granted and valid, under pending, and invalid. The legal status of patent families is more complex than that of patents, and there is no uniform definition so far. A "valid patent family" is generally defined as one of the members of a patent family whose legal status is granted and remains valid. In FRAND Rand licensing negotiations, there are times when the parties to the negotiation have a question about whether a patent family includes a specific country or region (e.g., China, the United States, Europe, Japan, and South Korea, i.e., China, the United States, Europe, Japan, and South Korea).Or only China, the United States, and Europe) are more interested. In this case, it is possible to define a specific valid family, such as "IP 5-office valid family" or "China-US-Europe valid family".
For the sake of simplicity, we use the example of "Chinese and American European families" with patent families in at least one of China, the United States, and Europe for statistical analysis. In this way, a "valid family in China, the United States, and Europe" can be defined as a patent family that has at least one patent member that has been granted and maintained in at least one of the offices of the Chinese Patent Office, the United States Patent Office, or the European Patent Office. Correspondingly, if the patent family of interest is only a patent family related to 5G technology, the "5G China, the United States, and Europe effective family" can be defined accordingly as a patent family with at least one 5G-related patent member that has been granted and maintained in at least one of the offices of the Chinese Patent Office, the United States Patent Office, and the European Patent Office.
According to the batch search of 209,688 5G patent entries in the patent database, 42,768 of the 53,670 5G patent families have patent families in at least one of the Chinese Patent Office, the US Patent Office, and the European Patent Office, that is, there are 42,768 5G "Chinese and American European families". Among the 42,768 5G EU, 32,195 have at least one authorized patent family in the Chinese Patent Office, the US Patent Office, and the European Patent Office, that is, there are 32,195 5G EU-EU valid families. According to the standardized patentee data, the number of 5G valid families owned by each patentee can be listed, and the distribution of the top 5 5G valid families in China, the United States and Europe and all other holders is shown in Table 1 below (the data in the table below is for example purposes only and is not real data).
Table 1 Statistics on the number of effective families in 5G, China, the United States, and Europe based on a special report by ETSI.
According to the results of this data analysis, the top 15 companies account for nearly 90% of all 5G effective families in China, the United States and Europe, and the top 5 companies have a share of more than 50%. In addition, it should be noted that the above statistics on the number of patent families for each patentee are determined based on the chronological order of the declaration documents from the nearest to the farthest, that is, if the same patent is claimed by different patentees, the patent is attributed to the patentee with the latest declaration. However, the actual situation is much more complicated. In the event of an assignment of a patent family, some assignees do not go to ETSI to re-declare, which will result in the relevant patent families still being attributed to the original patentee as the assignor for the time of statistics. This is further complicated in cases where patent families are assigned multiple times. In addition, some patent families only disclose some members of the transferred family at the time of assignment, which also brings some disagreement in determining the ownership of patent families. Therefore, in the process of real FRAND RAND licensing negotiation or litigation, the inventory of the number of SEPs also needs to be more accurately revised according to the actual situation taking into account a variety of factors (such as patent transfer, treatment of invalid and expired patents, etc.), especially for the confirmation of the number of SEP owned by both parties to the negotiation and litigation, and the latest patent transfer should be carefully verified to obtain the exact number of patent families.
Step 5: Necessity Analysis
As mentioned earlier, ETSI does not review the SEP necessity of the declaration. According to the ETSI IP Policy's definition of necessity, "necessity" applicable to IP rights means that it is not possible for technical, but not commercial, reasons why it is not possible to manufacture, sell, lease, otherwise dispose of, repair, use, or operate equipment or processes that comply with a standard without infringing that intellectual property right, taking into account the state of the technology generally available in normal technical practice and standardization.
As mentioned earlier, in the field of telecommunications, the general practice of FRAND licensing is to license globally or in multiple countries based on patent portfolios, which makes it necessary to count the patent families owned by SEP rights holders and judge the necessity of members within the patent family. In this case, it is necessary to distinguish between the standard necessity of a patent and the standard necessity of a patent family. A standard essential patent (i.e., SEP) is one that has at least one claim that corresponds to the relevant standard. A SEP family can be defined as a patent family in which at least one patent member is SEP, and correspondingly, a 5G SEP family can be defined as a 5G EUSEP family in which at least one active patent member is SEP in the Chinese Patent Office, the US Patent Office, and the European Patent Office.
When assessing the necessity of a single patent, in order to compare the patented technology with the standard text, it is necessary to evaluate whether the features in the claims of each patent are limited to specific standards, and whether the features are limited to the standards and require a skilled person in the relevant field or a lawyer with relevant professional knowledge to complete the standard. Therefore, it is a highly specialized and time-consuming, laborious and costly job.
In the pilot study of the European Commission's Joint Research Centre (JRC) on the necessity assessment of SEPs [4], the JRC provided the following table 2 on the time and cost of conducting the necessity assessment
Table 2 Costs of the need assessment for JRC statistics.
Based on the JRC's own necessity assessment pilot experiment, 28 evaluators, consisting of patent examiners, patent attorneys, senior engineers, and junior engineers under supervision, spent 1,414 hours to complete the necessity assessment of 205 patents, i.e., an average of 69. Evaluate the necessity of a patent. In addition, the JRC found that necessity itself is a binary concept (i.e., a patent is either SEP or not), but that the assessment of necessity is not a simple 0 and 1 exercise. Conceptually, patents are either necessary or not, and there is no such thing as a degree of "necessity". However, the actual assessment is a complex process, and there may be some grey areas in it.
Generally speaking, when the number of SEP-related patents to be analyzed is small, it is possible to analyze whether the claims of the patents correspond to the relevant technical standards one by one. This is more common in some SEP-related patent infringement lawsuits. For example, in Unwired Planet v Huawei[5], the English court found that two of Unwired Planet's patents (EP '744 and EP '818) were valid and necessary, two patents (EP '287 and EP '514) were invalid, and the trial of the other patent (EP '991) was delayed. It is only on the basis of examining the validity and necessity of the patent in question that the English court will decide on the FRAND licence rate.
However, when the number of SEPs to be analyzed is large, a complete analysis of whether all of the patentee's patents are SEPs becomes economically prohibitive and cannot be carried out (see the evaluation fees in Table 2). In this case, this can be done by sampling the patents (or patent families) to be analyzed.
For example, in TCL vIn the Ericsson case[6], in determining the total number of SEPs, Selna J took into account the necessity study report issued by the TCL side to determine the number of SEPs for the industry as a whole. The study uses a sample analysis to analyze the number of SEPs across the industry and the SEP data of the top 15 patent holders, including Ericsson. In the study, TCL's experts collated and analyzed the claims data from the ETSI database and summarized 11,469 patent families, from which 7,106 UE-related patent families were obtained by removing patent families that did not contain claims related to user equipment (UE). According to the declared criteria (2G, 3G and 4G) generations, the patent families held by the top 15 patent holders are randomly selected 1 3 for necessity analysis, and the necessity rate of all patent families held by each patent holder and all patent families of the relevant generation (e.g. 4G) is inferred based on the results of the sample analysis.
For the results of a sampling analysis, the error rate can be calculated at a given confidence level based on the formula. Specifically, when the confidence level is 90%, the absolute error rate d is calculated as:
When the confidence level is 95%, the absolute error rate d is calculated as:
When the confidence level is 99%, the absolute error rate d is calculated as:
Assuming that the 32,195 5G valid families in the United States and Europe are sampled at a 3% sampling ratio, the sample size obtained is 966. Assuming that the necessity rate is 40% after necessity assessment, then in the above formula, n=966, f=3%, p=04, so the calculated absolute error limit d is shown in Table 3 below
Table 3 Calculation of the error rate of sampling assessment.
As can be seen from Table 2, even if 32,195 5G U.S.-European-effective families are sampled at a very small sampling ratio of 3%, a high degree of confidence can be achieved with a small error rate.
And when the sample is relatively large, such as the TCL v. mentioned aboveWhen the sampling ratio in Ericsson reaches 1 3, the error rate becomes smaller.
When analyzing SEP patents, the evaluator needs to compare the content of the claims with the possible communication standards (e.g., the declared communication standards or the potential communication standards determined based on the keywords of the technical solutions of the claims), and the accuracy of the evaluation results will be improved if the evaluator can communicate with the inventor of the patent under evaluation or the contributor to the relevant standard technical proposal. In addition, in order to support the reproducibility of the analysis results, it is better for the analyst to fully document the results of the analysis with a claim comparison table or at least a key text commentary, rather than just giving a necessity assessment.
1.2 Intensity analysis based on the IPLYTICS database
The IPLYTICS platform claims to integrate and connect SEP declarations, standard documents, and technical proposals from global patent databases, standards organizations, and patent pools, and has consistently published rankings of 5G SEPs for many years.
To this end, we searched for 5G patents as of August 7, 2023 based on the IPLYTICS database, and compiled statistics from the perspective of valid patent families in China, the United States, and Europe, which are similarly defined as described above. The analysis process is based on the specific search of that day only. The data and analysis results in this example are not applicable to other areas, but the methodology used is somewhat universal.
The distribution of the top 5 5G holders in China, the United States, and Europe, as well as all other holders, is shown in Table 4 below
Table 4 Statistics on the number of effective families in China, the United States, and Europe for 5G based on IPLYTICS.
Compared with the data in Table 1, the top 15 patentees in this analysis are different, on the one hand, because of the difference in the statistical data, and secondly, because the deadline of the statistics is different, and the processes of data cleansing and family classification, query validity, etc. are not exactly the same, so they are all examples based only on the data of the current day, and the results are not applicable to subsequent situations.
In addition, it should be noted that when determining the ownership of a patent right, two or more patentees may apply for a joint application. In this case, the sum of the number of patents counted by individual patentees and the number of patents added together will be greater than the actual total number of patents, and the number of patents will account for more than 100%. In order to prevent this situation, and taking into account the relatively small proportion of joint applications, the sum of the number of patents is the sum of the number of patents listed in the previous section for the purposes of the above-mentioned statistics.
In addition to the above-mentioned methods of patent-by-case analysis and sampling, Tim Pohlmann, the founder of IPLYTICS, proposed a necessity assessment method based on an AI model [7], in which the necessity is assessed using easily observable patent features as variables related to potential semantic analysis results, third-party forward citations, size of patent families, inventors, priority dates, and technical proposals. In addition, it compares the model-based necessity assessment method with the evaluation methods of case-by-case analysis and sampling analysis, and believes that when the number of patents to be evaluated is small, the patent-by-patent analysis is the most accurate, but it has a large systematic error (the evaluator may be biased towards the patentee with an interest in it), so it may be beneficial to the accuracy of the results by fuzzying the patentee information of a specific patent and then providing it for evaluationWhen a smaller sample is drawn from a larger patent portfolio, the sampling analysis approach may not be as attractive as the necessity assessment method based on AI models.
Currently, the IPLYTICS platform provides a semantic necessity score that uses semantic analysis to assess the likelihood that a patent is necessary for the standard it claims [8], using a semantic mapping of the claims to the criteria section of a stand-alone patent to calculate a necessity score with data ranging from 1 to 100, where a numeric value of 10 indicates that the patent in question is most likely to be necessary for the standard. In addition, the platform integrates seven different patent evaluation metrics for evaluating patent portfolios, including technology relevance (TR), market coverage (MC), aggressiveness (RA), legal breadth (LB), patent scope (SC), collaboration (CO), and team size (TE) [9].
In the above-mentioned pilot study of the European Commission's JRC on the necessity assessment of SEPs, the JRC also studied the automation of the necessity assessment. They believe that automation solutions have the potential to be used as an auxiliary tool to improve efficiency and save resources, but they cannot replace human effort in the short term for a number of reasons: 1) the terms and interpretations in standards and patents depend on the context and may change with technological developments, making them difficult for machines to recognize;2) Although sometimes the description in the patent is different from the terms and actual embodiment in the standard, the patent may be necessary for the standard in substance;3) AI evaluation models need to have a sufficiently large reference dataset to be trained, which should include SEPs and non-SEPs, but such reference datasets do not currently exist, or even if some patent holders have such datasets, they are not willing to provide the portion that includes non-SEPs;and 4) even if such an AI evaluation model exists, it is unclear whether the patent model will be recognized by patent owners, standard implementers, courts, etc.
Therefore, the AI-based necessity assessment results provided by the IPLYTICS platform may only be used as a reference for enterprises to conduct general market assessments or understand industry trends, and may be questioned if they are used as bargaining chips or court evidence.
1.3 Outlook for the draft new EU SEP regulatory rules
On 27 April 2023, the European Commission published a draft "Proposal for a Regulation of the European Parliament and of the Council on Standard Essential" [CO(2023)232 - Proposal for a Regulation of the European Parliament and of the Council on Standard Essential patents and amending regulation (eu) 2017 1001][10]。 According to the description of the proposal, it aims to facilitate SEP licensing negotiations and reduce transaction costs for SEP holders and implementers by: (i) making it clearer who owns SEPs and which SEPs are really important;(ii) greater clarity on FRAND royalties and other terms and conditions, including increased awareness of value chain licensing;and (iii) facilitating dispute resolution for standard essential patents. To this end, the European Union Intellectual Property Office (EUIPO) will set up a Competence Center to manage SEP registration matters and patent necessity checks. If the draft is eventually implemented, the SEP assessment results provided by the CCC may become neutral and authoritative data for SEP strength analysis, which will undoubtedly provide more references for SEP rights holders and implementers in licensing negotiations.
2.Other strength analysis protocols
In addition to the intensity analysis methods and approaches introduced above, there are several ways to obtain SEP intensity (not limited to the communication field, but also the automotive field, image and coding fields, etc.).
One way is to leverage operational patent pool data. A patent evaluation is generally conducted before a standards-related patent pool is constructed to determine which patents are SEPs that can be placed in the pool. As a result, it is possible to analyze the data of patents using the operating patent pool to eliminate the need to evaluate the necessity of SEP, compared to the declared patents that need to be evaluated by the standards organization. However, operational pool data may be open to members of the pool, and most of the public usually only has access to information about the areas covered by patents, the total number of patents, etc., so it is difficult to apply them widely.
Another approach is to determine a search query based on the technology field to which the SEP belongs and apply the search query to the patent database to obtain a complete patent package to be analyzed. In this case, the accuracy and completeness of the search query determines the accuracy and completeness of the patent portfolio to be analyzed. For example, the Blue Paper Report on Wi-Fi 6 Technology published by the China Academy of Information and Communications Technology (CAICT) in 2022 [11] uses this method to determine patent collections, specifically, using Wi-Fi 6 name keywords, technology branch keywords, and Wi-Fi 6 standard keywords to construct search queriesSupplementary search is carried out based on the name of the subject of the patent application and the inventor, and denoising is carried out based on the patent content. Finally, the search results obtained by inpadoc family extension are used to obtain Wi-Fi 6 patent search data. Then, based on this retrieval data, it performs homing and necessity assessment.
3.Conclusion
As an indicator to measure the influence and value of a company's SEP core technology in related fields, SEP strength analysis not only helps enterprises evaluate the standard necessity of patents, but also helps investors, partners or competitors to understand a company's SEP strength, patent layout and technological innovation capabilities, so as to provide valuable reference for the company's decision-making on investment and business cooperation, negotiation and litigation. Based on the specific data of a specific date that is currently available**, this paper gives the sharing and discussion of the statistics of the number of SEPs and the assessment of necessity based on that specific date, which is only used as an example, but the method used is of some general applicability. In the future, it is expected that there will be more authoritative and reliable data bases that can be used for SEP intensity estimation in the future.
Comment
1] International Organization for Standardization Intellectual Property Policy and China's Reference.
2] China declares 5G standard essential patents180,000 items, ranking first in the world (Intellectual Property News).
3] intellectual property rights (iprs).
4] JRC Publication Repository - Pilot Study on Essentiality Assessment of SEPs, Table 22, No. 10verses 4 and 63 knots.
5] The Supreme Court of the United Kingdom against Unwired Planet v huawei, huawei v.Conversant as well as ZTE VConversant's Judgment, para. 20.
6] judge selna determines frand rate and enters contract-type injunction on etsi seps (tcl v. ericsson),7] ttps: precision and bias in the assessment of essentiality rates in firms’ portfolios of declared seps.
8] what is the iplytics semantic essentiality score?
9] patent valuation indicators
11] Global Wi-Fi6 Technology Innovation and SEP Analysis Report (2022).
(This article only represents the author's point of view and does not represent the position of intellectual property).
Cover** Intellectual property.