T 0679/14 (Turbo code interleaver/QUALCOMM) of 19.1.2017

Summary of Facts and Submissions

I. The appeal lies from the decision of the Examining Division to refuse European patent application No. 99962985.0, which was filed as international application PCT/US99/28580 and published as WO 00/35103, because the application did not meet the requirements of Article 56 EPC.

II. The Examining Division found that the subject-matter of the independent claims of a main request was not inventive over the following prior-art document:

D2: Clark, G.C., Bibb Cain, J.: "8.3. Interleaver Structures for Coded Systems", Error Correction Coding for Digital Communications, pages 345 to 352, Plenum Press, New York, 1981.

The same objections as raised for the main request were considered to apply with regard to claims 1, 9, 19 and 27 of a first auxiliary request, claims 1, 12 and 22 of a second auxiliary request, claims 1, 5, 8 and 10 of a third auxiliary request and claim 1 of a fifth auxiliary request. A fourth and a sixth auxiliary requests were not admitted under Rule 137(3) EPC.

III. This is the second appeal relating to the above mentioned application. In the first appeal proceedings resulting in decision T 17/07 of 20 December 2010 the competent board had come to the conclusion that the claims then on file, submitted with the letter faxed on 24 November 2010, fulfilled the requirements of Articles 84 and 123(2) EPC, and decided to remit the case to the Examining Division for further prosecution because novelty and inventive step of that request had not been assessed by the first instance.

IV. In the examination proceedings after the first appeal, a first summons to oral proceedings to take place on 18 January 2012 was sent on 17 October 2011. After that a new date for oral proceedings was fixed eight times. On three of those occasions, the oral proceedings were postponed in reaction to requests by the applicant on grounds related respectively to non-observance of the two-month notice for the summons, family vacation of the attorney and illness (see letters of 13 January 2012, 12 July 2012 and 14 March 2013). On two occasions they were cancelled by the Examining Division, with the statements "oral proceedings cannot take place" and "the first member has to attend another meeting" respectively (see brief communications of 23 March 2012 and 11 January 2013). For the other three postponements, no reason is apparent from the file.

V. In the statement of grounds of appeal, the appellant requested that the contested decision be set aside and that a patent be granted on the basis of the main request filed with letter of 24 November 2010, corresponding to that on which the first appeal decision T 17/07 was based, or, alternatively, the first auxiliary request filed with letter of 16 December 2011, both requests resubmitted with the grounds of appeal.

VI. The appellant was invited to oral proceedings. In a subsequent communication sent in advance of the oral proceedings, the Board also referred to document D1, which had been cited by the Examining Division in its communication of 19 August 2002:

D1: Dolinar, S., Divsalar, D.: "Weight Distributions for Turbo Codes Using Random and Nonrandom Permutations", TDA Progress Report 42-122, 15 August 1995.

The Board drew attention to apparent deficiencies, to be discussed at the oral proceedings, of the examination proceedings before the first instance and of the reasoning of the contested decision.

The Board stated that, taking into account the long duration of the proceedings thus far, the fact that this was the second appeal in the present case, and that the Board had to assume that a proper search had been carried out, it was not inclined to remit the case to the department of first instance for further prosecution regardless of whether the deficiencies were found to amount to procedural violations.

The Board made a preliminary assessment of inventive step of the subject-matter of the independent claims of the main request, taking document D2 as the starting point. It also summarised the disclosure of document D1, which it considered also relevant for the assessment of inventive step. With regard to the auxiliary request, the additional feature seemed to be well known and would hence not change the result of the assessment of inventive step compared with that of the main request.

VII. With a letter of reply the appellant filed a second auxiliary request.

VIII. Oral proceedings were held on 19 January 2017. During the oral proceedings the appellant filed a new set of claims as sole request, replacing all previous requests, and a complete set of pages of the description, and requested reimbursement of the appeal fee. At the end of the oral proceedings, the chairman pronounced the Board's decision.

IX. The appellant's final requests were that

- the contested decision be set aside;

- the case be remitted to the department of first instance with the order to grant a patent on the basis of claims 1 to 38 and description pages 1 to 3, 3a, 4, 4a and 5 to 21 as filed during the oral proceedings on 19 January 2017, and of the drawings of Figures 1 to 3 as published;

- the appeal fee be reimbursed.

X. Claim 1 of the sole request reads as follows:

"An interleaver (16) for a turbo coder (10), comprising:

means for writing data elements sequentially by rows to a matrix of bit storage locations, the matrix comprising a first plurality of rows and a second plurality of columns;

means for permuting the rows of the matrix;

means for pseudo-randomly rearranging the data elements in each row in the matrix of bit storage locations in accordance with a linear congruential sequence recursion, wherein the linear congruential sequences associated with two distinct rows are different; and

means for reading the data elements sequentially by columns from the matrix of bit storage locations."

Independent claim 10 reads as follows:

"A turbo coder (10), comprising:

a first coder (12) configured to receive a plurality of input bits in succession and generate a first plurality of output symbols therefrom;

an interleaver (16), comprising:

means for writing the plurality of input bits sequentially by rows to a matrix of bit storage locations, the matrix comprising a first plurality of rows and a second plurality of columns;

means for permuting the rows of the matrix;

means for pseudo-randomly rearranging the bits in each row in the matrix of bit storage locations in accordance with a linear congruential sequence recursion, where in [sic] the linear congruential sequences associated with two distinct rows are different; and

means for reading the bits sequentially by columns from the matrix of bits storage locations to provide a plurality of interleaved bits; and

a second coder (14) configured to receive the plurality of interleaved bits in succession from the interleaver (16) and generate a second plurality of output symbols therefrom."

Independent claim 21 recites "A method of interleaving data elements for a turbo coder (10)" comprising the four steps defined in claim 1 with respect to the means of the interleaver.

Independent claim 30 recites "A method of turbo coding" comprising the steps defined in claim 10 with reference to the components of the turbo coder.

Claims 2 to 9, 11 to 20, 22 to 29 and 31 to 38 are dependent claims.

XI. Insofar as relevant for the present decision, the appellant's arguments were as follows:

The present invention provided for an interleaver, in particular a code interleaver of a turbo coder, with better performance and reduced implementation complexity compared to other existing known turbo code interleavers.

The interleaver was essentially a block, or row-column, interleaver with both row-interleaving and intra-row elements permutation, the latter being done in linear congruent form. A turbo code could be improved by using such an interleaver "inasmuch as it was possible to avoid or at least minimise bad mappings of error events".

The Examining Division appeared not to have taken into account the advantageous aspects of the invention, in particular better performance of the interleaver and reduced implementation complexity. Document D2 only related to a single sequence of integers and did not contemplate any addressing supporting a mathematical construct of rows and columns. Such a construct of a two-dimensional interleaver employed by a parallel turbo coder generally outperformed a parallel turbo coder having a one-dimensional interleaver (such as the one described in D2) in terms of coding gain.

The Examining Division appeared to be of the opinion that rows and columns were merely "metaphorical" but not real. However, whether physical or mathematical, the indexing of rows and columns was indeed real and significant in the present invention, and could not be mischaracterised as, or reduced to, a metaphor for the sake of expediency.

Document D2 did not actually disclose or point to a two-dimensional interleaver in which rows were permuted and data elements in each row of the matrix were then pseudo-randomly rearranged in the matrix of bit storage locations using linear congruential sequences, wherein the linear congruential sequences associated with two distinct rows were different.

There was no teaching in the prior art to pick something from one embodiment and combine it with another embodiment.

Even taking into account the disclosure of document D2 and the concept of two-dimensional interleavers, there was no hint for the skilled person to use linear congruential sequences for element permutations in individual rows, wherein the linear congruential sequences associated with two distinct rows were different. Moreover, in D2 there was no hint that these rows were permuted. Even if the skilled person combined the different embodiments of document D2, it would not be obvious to arrive at the claimed solution. The skilled person would not additionally permute rows.

The first embodiment of document D1 mentioned in the Board's communication already proposed a solution to improve interleaving based on column permutations in addition to row permutations. The skilled person would therefore not search for another solution to the problem. There was no hint in the prior art to modify the interleaver of D1 by replacing those column permutations with different linear congruential sequences in distinct rows.

The appellant was dissatisfied with the way the case had been handled in the first-instance proceedings. The repeated adjournment of oral proceedings had been prejudicial to it. The reasoning of the Examining Division was unclear. The appellant was not in favour of remittal to the department of first instance for assessment of inventive step.

Reasons for the Decision

1. The appeal complies with the provisions referred to in Rule 101 EPC and is therefore admissible.

Appeal - procedural aspects

2. Decision T 17/07 of the first appeal in the present case established that the claims then on file satisfied the requirements of Articles 84 and 123(2) EPC.

The claims now on file differ from those on which that first appeal decision was based in that "for a turbo coder (10)" has been added to the text of independent claims 1 and 21. Since that amendment does not prejudice clarity and finds direct and unambiguous support in the application as originally filed (see e.g. the title or the brief description of Figure 2 on page 4, lines 33 and 34, of the present application), it does not affect the conclusion of the first appeal decision with regard to clarity and added subject-matter.

The finding of that first appeal decision being res judicata (see decision T 153/93 of 21 February 1994, reasons 2 and 3, Case Law of the Boards of Appeal, 8th edition 2016, IV.E.7.7 and 7.7.4, and T 449/15 of 25 February 2016, reasons 2), the Board confirms that the claims satisfy the requirements of Articles 84 and 123(2) EPC.

3. At the oral proceedings the appellant expressed its dissatisfaction with the way the case had been handled in the first-instance proceedings and said that it found the reasoning of the Examining Division unclear. Nevertheless, it was not in favour of remittal to the department of first instance for assessment of inventive step and accordingly requested further prosecution by the Board.

4. This is the second appeal in the present application, which entered the European phase in June 2001. In the Board's view, the long duration of the proceedings thus far and the fact that this is the second appeal constitute special reasons under Article 11 RPBA for not remitting the case in reaction to fundamental deficiencies in the first-instance proceedings. The Board therefore considers that, irrespective of its conclusions on the deficiencies found, it should further examine the claims with regard to inventive step.

Inventive step

5. Each of independent claims 1, 10, 21 and 30 recites the concrete application of the claimed invention for turbo coders. According to established case law, processes for encoding/decoding are to be regarded as technical processes, even if they are based essentially on mathematical processes (see also Case Law of the Boards of Appeal, 8th edition 2016, I.A.2.2.2). Interleaving is used to create a more uniform distribution of errors and improve the performance of the encoding and decoding process. It therefore improves the performance of a process viewed as technical. As a consequence, and although the interleaving method is based on mathematical operations, the Board recognises the technicality of the claimed invention.

6. Claim 1 recites an interleaver comprising means for writing data elements in storage locations, rearranging data elements and reading data from the storage locations. According to decision T 17/07, the claim "covers both means that physically write the permuted rows and rearranged data elements in the matrix and means that generate indexes which can be used for permuting the rows and rearranging the data elements in each row" (see reasons 3.1).

The fact that according to decision T 17/07 the claim covers two implementations of the interleaving method does not imply that the matrix features should be considered merely conceptual and ignored in the assessment of inventive step, as seems to have been done in the appealed decision at least with regard to some matrix features (see also point 14.614.6 below).

In each of the interpretations of T 17/07, the claimed interleaver has the effect that data retrieved from the storage locations is rearranged in relation to the data that was previously stored in those storage locations. The same applies to the corresponding subject-matter of the other independent claims.

In addition, since the claimed interleaver writes the data elements by rows but reads the data elements by columns rather than by rows, in addition to the other features of the claim, the matrix features cannot be simply mapped to features of an interleaver permuting data elements only within each individual subsequence of the received sequence of data elements, each subsequence corresponding to a row or a column. Insofar as the matrix features play a role in the interleaving performed and its associated properties, they should not be ignored in the assessment of inventive step.

7. Prior-art document D2 discloses interleaving techniques for coding systems (see e.g. page 345, first two text paragraphs, Figure 8-9). Even though it does not mention turbo coders, the Board considers that the skilled person would also take its teaching into account in the general context of interleavers for turbo coders. It is therefore an appropriate starting point for the assessment of inventive step in the present case.

7.1 Document D2 explains on page 345, second text paragraph, that two principal classes of interleavers are periodic and pseudorandom. Starting on page 346, in section 8.3.1 document D2 describes periodic interleavers, in particular two types: block interleavers and convolutional interleavers. Block interleavers are said to accept blocks and perform identical permutations over each block of symbols (page 346, first text paragraph). As an example of a typical block interleaver, document D2 discloses the "(B, N) block interleaver" in subsection 8.3.1.1, which takes the coded symbols and writes them by columns into a matrix of N rows and B columns, the permutation consisting of "reading these symbols out of the matrix by rows prior to transmission" (see page 346, second text paragraph). This appears to correspond to the means for writing data elements sequentially by rows to a matrix and reading sequentially by columns of the claim (interchanging writing by columns and reading by rows with writing by rows and reading by columns does not result in a different technical teaching).

7.2 The interleaver of claim 1 differs from the (B,N) block interleaver of document D2 in that it is an interleaver for a turbo coder and includes means for respectively ("rows" in the claim corresponding to "columns" in D2)

(a) permuting the rows of the matrix;

(b) pseudo-randomly rearranging the data elements in each row in the matrix of bit storage locations in accordance with a linear congruential sequence (LCS) recursion, wherein the LCSs associated with two distinct rows are different.

7.3 With regard to the problem solved by the claimed invention, in the grounds of appeal the appellant reiterated some of the arguments presented in the first-instance proceedings, namely that the claimed invention provided good error rate performance and simplicity of implementation (see also page 3, lines 19 to 37 of the description). At the oral proceedings the appellant added that the implementation of the receiving side in the system of the invention was simpler because it required only the parameters used in the interleaver.

The Board is not convinced that the distinguishing features result in reduced complexity of the implementation with regard to the (B,N) block interleaver of document D2, especially since claim 1 does not give any implementation details.

The contested decision did not provide any comments on the error rate performance. As explained in document D2, the (B,N) block interleaver lacks robustness if there is substantial variation in the characteristics of the burst noise process (page 346, penultimate paragraph).

The Board therefore recognises that the distinguishing features improve randomness and may potentially improve the error rate performance of the interleaver for some burst error patterns.

7.4 In the Board's view, the skilled person facing the above problem of improving the error rate performance would search for other permutation and interleaving techniques. However, none of the available prior-art documents discloses both features (a) and (b) used in the same context as in the present invention.

7.4.1 Document D2 discloses pseudorandom interleavers in section 8.3.2, starting on page 349. According to that disclosure, a pseudorandom interleaver is a block interleaver which takes a block of L channel symbols and reorders, or permutes, them in a pseudorandom fashion. Document D2 explains that if the same permutation is used on each interleaver block, certain interference patterns will exist that can seriously degrade the performance. It then teaches changing the permutation frequently to avoid that problem, e.g. by storing a fixed number of permutations, say M, in a ROM and randomly selecting a permutation from this group for each interleaved block. An example of such an interleaver is described in the paragraph bridging pages 349 and 350 with reference to Figure 8-11 of page 350. On page 350, document D2 discusses the question of selecting the permutations to be used and refers to the use of LCSs for the permutation (see page 350, Equation (8-2), and preceding paragraph). The equation expressing the pseudorandom sequence in document D2, Equation (8-2), actually corresponds to the equation of claim 3 of the present application, which further defines the LCS of claim 1. These features of a pseudoramdom interleaver with a new permutation for each succeeding interleaved block correspond to those of distinguishing feature (b).

At the oral proceedings the appellant argued that the pseudorandom interleavers disclosed in section 8.3.2 of document D2 did not use a matrix; interleaving was performed only on one column. The Board agrees but notes that the matrix features are already disclosed for the (B,N) block interleaver of section 8.3.1.1. The question is therefore whether the skilled person would enhance that (B,N) block interleaver by adding the pseudorandom interleaving of section 8.3.2 corresponding to feature (b) above and thereby arrive at the claimed solution. With respect to that question, the Board agrees with the appellant that none of the documents hints at using LCS permutations in individual rows of a (B,N) block interleaver, and that the combination of the two separate embodiments of D2 would still not include feature (a). The Board is therefore persuaded that it would not be obvious for the skilled person to arrive at the claimed invention when starting from document D2.

8. Document D1 discloses in the context of turbo coding block interleavers defined by a matrix (see page 61, section B.1), in which columns are reordered. It teaches that a block interleaver can be improved by permuting rows in addition to its columns (page 61, last four lines). That embodiment therefore does not include feature (b) either.

On page 62, document D1 discloses permutations based on circular shifting similar to those of feature (b). However, those features are disclosed with regard to a separate embodiment (see page 62, section B.2). At the oral proceedings, the appellant argued that in the light of the disclosure of the solution including both row and column permutations for the first embodiment of document D1 (page 61, section B.1), the skilled person would have no incentive to remove from that embodiment the column permutations and replace them with the circular shifting of the second embodiment applied at each row. The Board agrees with the appellant, noting that the idea of performing different permutations in distinct rows is clearly different from that of permuting columns across all rows of the matrix. None of the available prior-art documents discloses how to combine the circular shifting with a block interleaver like that disclosed in section B.1 of document D1.

9. The Board is therefore satisfied that the cited prior-art documents, either individually or in combination, do not render obvious the subject-matter of claim 1, the same applying for corresponding independent claims 10, 21 and 30. The application therefore fulfils the requirements of Articles 52(1) and 56 EPC.

Remittal

10. From the above the Board concludes that the claims comply with Articles 84, 123(2), 52(1) and 56 EPC. The description has been adapted to the present claims. No other deficiencies can be identified with regard to the application documents on file. The Board therefore decides to set aside the decision under appeal and to remit the case to the department of first instance with the order to grant a patent on the basis of the sole request on file.

Reimbursement of the appeal fee

11. The appellant's request for reimbursement of the appeal fee can be granted only if such reimbursement is equitable by reason of a substantial procedural violation (Rule 103(1)(a) EPC).

12. The deficiencies apparent in the first-instance proceedings in the present case, which were discussed in the Board's communication and at the oral proceedings, concern, on the one hand, the multiple postponements of oral proceedings by the Examining Division and, on the other hand, the written decision.

13. Postponements of oral proceedings by the Division

13.1 It is contrary to the principles of procedural efficiency and legal certainty to repeatedly adjourn oral proceedings. According to the Guidelines for Examination in the EPO, E-II, 7 and the case law cited there, a request for postponement of oral proceedings is allowable only if the party concerned can advance serious reasons which justify the fixing of a new date. Regarding postponements at the instigation of the Division, the Guidelines, E-II, 7.1, state that in exceptional cases the Division might have to postpone oral proceedings for reasons similar to those mentioned for the parties but only if a suitable replacement cannot be found.

13.2 The Board finds, in line with the Guidelines, that it is unacceptable to repeatedly postpone oral proceedings without serious reasons, especially if it is done several times in the same examination proceedings after long delays have already occurred. Fixing a new date for oral proceedings almost inevitably causes unnecessary work and additional costs, also for the attorney and/or applicant, for example due to re-planning, cancelling a booked business trip, making new travel arrangements, or repeated preparation of the case. Furthermore, the negative consequences of repeated delays are not limited to the applicant because legal certainty is necessary to protect the interests not only of the applicant but also those of the general public or third parties.

13.3 In the present case, the application entered the European phase in June 2001. The eight postponements of oral proceedings (see section IV above) caused an extra delay of almost one and a half years after more than ten years of examination (including the first appeal). Two postponements were at the instigation of the Examining Division and not based on serious reasons. The same can be assumed to apply to the other three postponements for which no request by the applicant and no other ground for postponement can be found in the file. In the opinion of the Board, the first member's attendance at a meeting does not in principle qualify as a serious reason. Repeated fixing of new dates for oral proceedings could have been avoided by replacing one or more members of the Examining Division, as indicated in the Guidelines.

The Board finds that in the present case, although the individual delays caused by the postponements were short, the Examining Division acted against the interests of procedural efficiency, at the same time ignoring clear rules given in the Guidelines to avoid such delays.

At the oral proceedings before the Board, the appellant explained that during the first-instance proceedings the applicant had expressed its disagreement with the repeated rescheduling of oral proceedings, and that the postponements had caused additional overheads and expenses. It is however not apparent from the file to which extent the applicant made such submissions and firmly objected to the postponements during the first-instance proceedings.

13.4 Taking the above into account, the Board concludes that the postponement of oral proceedings on those five occasions at the instigation of the Examining Division without serious reasons, as far as apparent from the file, constitute a procedural deficiency in the circumstances of the present case. The Board nevertheless refrains from deciding whether it amounts to a substantial procedural violation making the reimbursement of the appeal fee equitable within the meaning of Rule 103(1)(a) EPC.

14. Written decision

14.1 According to Rule 111(2) EPC, decisions of the EPO open to appeal must be reasoned. The case law of the Boards of Appeal establishes criteria for substantiating those decisions (see e.g. Case Law of the Boards of Appeal, 8th edition 2016, III.K.4.2.1), for instance:

"The reasoning given in a decision open to appeal has to enable the appellants and the board of appeal to examine whether the decision was justified or not. A decision therefore should discuss the facts, evidence and arguments which are essential to the decision in detail. It has to contain the logical chain of reasoning which led to the relevant conclusion".

The reasoning and grounds should "be comprehensible to those conducting a later judicial review". According to decision T 1123/04 of 25 August 2006, "it should not be necessary for a board of appeal to have to reconstruct or even speculate as to the possible reasons for a negative decision in the first instance proceedings." A decision should be complete and self-contained (see reasons 3.3).

14.2 Even though the decision under appeal does not properly identify at least some of the claims on which the decision was based, after an analysis of its grounds it can be concluded that the reasoning applies to the requests on which the decision should have been based, namely the main request considered in decision T 17/07, and the first to sixth auxiliary requests filed with letter dated 16 December 2011.

14.3 In the decision under appeal the Examining Division did not explicitly mention what it considered to be the starting point for its inventive-step argumentation. Since it compared the invention with the disclosure of Figure 8-11 and Equation (8-2) of document D2, the Board concludes that it considered document D2 to be the closest prior art. However, the contested decision referred only to Figure 8-11 and Equation (8-2) of document D2, without clearly explaining how they corresponded to claimed features or their relevance for the assessment of inventive step.

In particular, the Examining Division stated that "figure 8-11 of D2 implements a linear recursive sequence interleaving rule; see equation 8-2 of D2". However, Figure 8-11 and Equation (8-2) cannot be understood without a closer reading of text passages of the document. Furthermore, there is no reference in Figure 8-11 to Equation (8-2) or vice versa. It cannot be concluded from their content alone that they are disclosed in combination in document D2. In order to be understandable, the decision should therefore have explained how the Examining Division interpreted the cited figure and equation and should have cited text passages of document D2 supporting its interpretation.

14.4 A passage of section 2 of the decision relating to the assessment of inventive step of claim 1 of the main request reads as follows:

"Construing claim 1 of the main request as described above, i.e. following T17/07, the question is how the subject-matter of claim 1 differs from the interleaving method of D2, figure 8-11. Claim 1 allows that the frame length L is the same as in D2, which means that according to claim 1 this frame length is divided into plural segments, with then a LCS interleaving being applied to each segment. It is not clear what the effect of this difference should be on the interleaving properties such as interleaving distance. As claim 1 does not specify the number of segments, this number can be high, e.g. 16 segments for L=32, which means two symbols per segment. Obviously, in this case the interleaving is very small, as the symbols are then only permuted

inside their own segment."

In this passage the Examining Division refers to "this difference" without clearly saying what the difference to the prior art is. The Board speculates that the difference is that "this frame length is divided into plural segments, with then a LCS interleaving being applied to each segment", but the claim mentions neither frames nor segments. It is not clear either what "frame length L" is with respect to the claimed interleaver. It seems from page 5, last paragraph, that the Examining Division considered the difference to reside in LCS interleaving being applied to segments instead of to the whole frame (whatever the frame and segments mean in the context of the claim), and it could be that segments were considered to correspond to rows. In any case, the reasoning for claim 1 of the main request does not further clarify those points and does not discuss the features related to permutations of rows, writing by rows and reading by columns.

14.5 In the opinion of the Board, the reasoning with respect to the auxiliary requests, including non-admission of the fourth and sixth auxiliary requests under Rule 137(3) EPC, is also incomplete and unclear. With regard to the fourth auxiliary request, the reasoning given in the contested decision was that the independent claims no longer defined that the interleaving was done according to linear recursive sequences. For the sixth auxiliary request it was essentially explained that claim 1 was related to details of Figure 2, but that the generalisation to an interleaver controller as opposed to an interleaver was not admitted under Rule 137(3) EPC. Without further reasons, these are not valid arguments for not admitting those claim sets into the proceedings under Rule 137(3) EPC.

14.6 In the grounds of appeal the appellant complained that the decision mischaracterised the features describing the indexing of rows and columns and reduced them to a metaphor for the sake of expediency.

In the Examining Division's communication accompanying the summons to oral proceedings the notion of rows and columns had been seen as being "only metaphorical". In the decision under appeal, the Examining Division no longer referred to these features as "metaphorical" but stated, with regard to the fifth auxiliary request and in section 9 (see page 7), that the language of the claim, in particular the description of interleaving in terms of a matrix having rows and columns with permutations being done inside the matrix, was merely a conceptual description, not a reference to real technical (i.e. structural or electronic) features of the interleaver.

It is not clear whether the Examining Division's view that the "matrix features" were merely conceptual also applied to the main request and first to third auxiliary requests.

Since some of the matrix features e.g. columns, writing by rows and reading by columns, have not been discussed in the assessment of inventive step, the Examining Division seems to have considered that those features lacked technical character and therefore did not have to be taken into account. On the other hand, assuming, as explained under point 14.414.4 above, that the Examining Division mapped rows of the claimed interleaver to segments in document D2, at least one of the matrix features was not ignored in the assessment of inventive step. Similarly, for the first auxiliary request the Examining Division referred to "the row permutations to increase data shuffling". It is then unclear why some but not all "merely conceptual" matrix features should be ignored.

14.7 With regard to the inventive-step reasoning for the first auxiliary request, the Examining Division first briefly referred to the interpretation of the feature "bit reversal rule" of claim 1. It afterwards stated the following:

"Also in this case it is not clear what the technical effect should be compared to D2, in which one LCS permutation is applied to the whole input data frame. It is understood that according to the application this row permutation is to increase the data shuffling compared to not having this row permutation. However, the LCS interleaving of D2 was not acknowledged in the application as filed, so that it is not clear what the technical effect of the interleaver of claim 1 - interpreted according to T 17/07 - compared to that of D2 should be" (see paragraph bridging pages 5 and 6 of the contested decision).

The Examining Division then concluded that a technical inventive concept could therefore not be recognised, contrary to Article 56 EPC.

Stating that it is not clear what the technical effect is because document D2 was not acknowledged in the application as filed seems to imply that in order to have the technical effect of its invention recognised over the closest prior art the applicant would have to acknowledge the closest prior art in the original application. This contradicts the case law regarding acknowledgement of prior art in patent applications as originally filed. In particular, according to decision T 2321/08 of 11 May 2009 (followed by T 1123/09 of 17 December 2009, which was cited in the communication accompanying the summons to oral proceedings), "Rule 27(1)(b) EPC 1973 does not put a stringent obligation on the applicant to cite documents reflecting prior art known to him already at the time of filing the application" (see reasons 7.3) and the applicant may amend the application to acknowledge prior art during examination (reasons 8). Logically, the applicant is not obliged to acknowledge prior art unknown to it either. Similarly, as explained in decision T 105/11 of 10 March 2016, "Rule 42(1)(c) EPC merely requires the invention to be presented in such terms that the technical problem and its solution can be understood" (see reasons 4.9 and 4.10).

Furthermore, from the reasoning above it is not clear whether the Examining Division conceded that the "data shuffling" was considered to be increased and why. If it accepted increased "data shuffling", it is unclear why it could not recognise any associated technical problem in the context of interleaving.

In the Board's view, it is the duty of the Examining Division to identify the objective technical problem and technical effect of the distinguishing features with regard to the closest prior art or, alternatively, to clearly explain why the distinguishing features do not solve a technical problem, e.g. because they do not have a technical effect. The above reasoning does not fulfil either of those prerequisites.

14.8 In the grounds of appeal, the appellant argued that the Examining Division appeared not to have taken into account the advantageous aspects of the invention, in particular reduced implementation complexity and better performance of the interleaver. The Board finds that the decision under appeal discusses the first alleged advantage of implementation simplicity but agrees with the appellant that it does not address the second advantage mentioned by the appellant.

Indeed, the Board could not find in the contested decision a discussion of the applicant's argument that the claimed interleaver resulted in a good error rate performance. This argument was invoked in the applicant's letter of 11 May 2012 (page 2, first three lines), was repeated in the oral proceedings (see minutes, page 2, paragraph 8), and seems to correspond to the "coding gain" advantage mentioned on page 3, lines 19 to 33 of the application. Furthermore, good error rate performance could be considered to be related to the increased "data shuffling" mentioned but not further discussed in the appealed decision. That argument of the applicant was therefore relevant for the decision and should have been addressed.

14.9 For the reasons given above, the Board shares the appellant's view that the reasoning of the contested decision is unclear.

The decision under appeal does not deal with all the features of claim 1 of the main request and of the other refused claims (see e.g. points 14.414.4 and 14.614.6 above), does not provide a logical sequence of arguments that justify its tenor (see points 14.314.3 to 14.7), fails to address an essential argument of the applicant (see 14.8 above) and does not discuss in detail some of the facts, evidence and arguments which are essential to the decision (e.g. other relevant passages of document D2, see point 14.314.3 above).

Consequently, the appealed decision is insufficiently reasoned and does not meet the requirements of Rule 111(2) EPC, which constitutes a substantial procedural violation. As the appeal is deemed to be allowable and reimbursement of the appeal fee is equitable by reason of a substantial procedural violation, the request for reimbursement of the appeal fee is to be granted in accordance with Rule 103(1)(a) EPC.

Order

For these reasons it is decided that:

1. The decision under appeal is set aside.

2. The case is remitted to the department of first instance with the order to grant a patent on the basis of the following documents:

- claims 1 to 38 as filed during the oral proceedings of 19 January 2017;

- description pages 1-3, 3a, 4, 4a, 5-21 as filed during the oral proceedings of 19 January 2017;

- drawings of Figures 1 to 3 as published.

3. The appeal fee is to be reimbursed.