T 2177/13 (Decoding hypotheses/QUALCOMM) 02-12-2016

1. Non-attendance of the appellant at oral proceedings

1.1 The appellant decided not to attend the scheduled oral proceedings before the board (cf. point VII above). Pursuant to Article 15(3) RPBA, the board is not "obliged to delay any step in the proceedings, including its decision, by reason only of the absence at the oral proceedings of any party duly summoned who may then be treated as relying only on its written case."

1.2 In the present case, the appellant submitted comments in support of the patentability of the claim requests on file in response to the objections raised in the board's communication under Article 15(1) RPBA (cf. point VI above). The board considered those comments but maintained its objections under Article 56 EPC (cf. points 2.1 and 3.1 below). So, in the exercise of its discretion under Article 15(3) RPBA, the board took a decision at the end of the oral proceedings, in the absence of the duly summoned appellant.

2. MAIN REQUEST

Claim 1 of the main request is identical to claim 1 underlying the appealed decision.

2.1 Novelty and inventive step (Articles 54 and 56 EPC)

2.1.1 The board first notes that it understands the term "bit value of the encoded data bit" as used in present claim 1 to refer to "any physical or logical value associated with a received data bit".

2.1.2 As regards the closest prior art for the subject-matter claimed, the board considers prior-art document D6, introduced by the board, to be a more suitable starting point for assessing novelty and inventive step than documents D1 and D5, cited in the decision under appeal (cf. point I above). Document D1 relates to Viterbi-decoding of encoded data based on predetermined synchronisation bits and bit locations but is silent about generating and processing multiple hypotheses based on those synchronisation bits. Document D5 is concerned with data decoding relying on cyclic redundancy checks (CRCs) using multiple sampling methods to sample the received bits in parallel, prior to decoding and selecting the "best" signal from the sampled ones, and is likewise silent as to generating and processing multiple hypotheses based on different a-priori sample or CRC information.

2.1.3 The board holds that D6 discloses the following limiting features of present claim 1:

A method for decoding encoded data bits ("received bits vk" with bit positions "k") of a wireless communication transmission (see e.g. paragraph [0032]), comprising the steps of:

generating multiple hypotheses ("hypotheses h"), each hypothesis specifying a set of bit values of the encoded data bits ("sequence Sh of scaling factors sk") that are predicted based on a-priori information (e.g. "amplitude Ak"; "variance sigmak**(2)" of the received bits; see paragraph [0052], first sentence) regarding the transmission (see e.g. paragraph [0055]: "... a number of hypotheses are formed for a given code segment ... to be decoded. Each hypothesis h corresponds to a hypothesized sequence Sh of scaling factors ...", in conjunction with paragraph [0009]: "... The scale information may include the variance and amplitude of the samples ..." and Fig. 6, step 612);

(b) evaluating the multiple hypotheses by performing decoding for the encoded data bits (see e.g. paragraph [0057]: "... the code segment is Turbo decoded in accordance with the selected hypothesis ... to provide a candidate sequence of decoded bits ..." in conjunction with Fig. 6, steps 614 to 618);

(c) selecting, as output, decoded bits corresponding to one of the hypotheses (see e.g. paragraph [0013]: "... The quality of the decoded results for all hypotheses are compared, and the decoded bits for the hypothesis that provides the 'best' decoded result ... are provided as the Turbo decoder output for the code segment", in conjunction with paragraph [0058] and Fig. 6, steps 620 to 626).

2.1.4 As to feature (a), the appellant persistently argued that the "hypotheses" used in document D6 related to scaling factors, i.e. to attributes of a soft bit, rather than to a "bit value of an encoded data bit" such as "0" or "1" as defined in present claim 1 (cf. appellant's letter dated 31 May 2016, page 5). Thus, according to claim 1, the transmitted information itself was hypothesised and not the manner in which the information was represented (cf. appellant's letter dated 2 November 2016, pages 2 to 4).

The board is not persuaded by this argument. Firstly, the term "bit value" is not necessarily limited to the logical values of a data bit like "0" and "1" (cf. point 2.1.1 above). Secondly, even if the notion "bit value" specified only logical values, it is apparent to the board that D6 in fact indicates that each hypothesis h corresponds to "a particular set of one or more values for a set of one or more parameters used for decoding the code segment" (see paragraph [0012], second sentence) and that "scale information may relate to the variance and amplitude of the bits to be decoded, and may be needed to accurately evaluate a particular function for the MAP decoding" (see paragraph [0011], third sentence). From this the skilled reader may immediately infer that by means of the scale information the actual logical bit values ("1" or "0") of the encoded data bits could readily be determined, depending on the scaling of the corresponding data bits. Hence, contrary to the appellant's assertions, those hypotheses of D6 indeed specify - via the scaling information - "a set of bit values of the encoded data bits" according to feature (a) of present claim 1. Thirdly, given that the scale information in D6 may even include the variance sigmak**(2) and amplitude Ak of the received bits (see page 2, penultimate sentence) and that each received bit vk can be calculated directly from sigmak and Ak (see page 7, equation Eq(2) in paragraph [0032]), the skilled reader of D6 would readily appreciate that the hypotheses h relating to the scale information sk indeed imply the hypothesised logical values of the encoded data bits.

2.1.5 However, the appellant was able to persuade the board that D6 fails to disclose directly and unambiguously that the decoding according to feature (b) is indeed performed by eliminating the sets of decoded bits which are inconsistent with those specified by the respective hypotheses.

2.1.6 Hence, the subject-matter of claim 1 differs from the disclosure of D6 in that

(b') the decoding involves eliminating sets of decoded bits that are inconsistent with the bit values specified by the hypotheses.

Accordingly, D6 does not take away the novelty of the subject-matter claimed (Article 54 EPC).

2.1.7 As regards the issue of inventive step, the skilled reader would in particular notice from the teaching of D6 that, by successive decoding of encoded data bits on the basis of the accordingly selected hypotheses, "candidate sequences of decoded bits" are established for the purpose of subsequently finding the "best hypothesis" (see e.g. paragraphs [0057] to [0059], in conjunction with Fig. 6, steps 614 to 622). The board holds that the skilled person in the field of channel coding, in order to solve the objective problem of "how to actually implement the selection of the candidate sequences in the decoding system of D6", would be well aware that sequences of decoded bits which do not match - for whatever reason - with bit values specified by the respective hypotheses, i.e. "non-candidate sequences", cannot be taken into account and therefore have compellingly to be eliminated somehow in the underlying decoding process. Consequently, in the absence of more specific measures regarding the consistency check and data elimination algorithm, distinguishing feature (b') constitutes a straightforward implementation detail for the skilled person and cannot confer inventive step on present claim 1.

2.2 In view of the above, the main request is not allowable under Article 56 EPC.

3. AUXILIARY REQUESTS

Claim 1 of the first to fourth auxiliary requests further specify that (emphasis added by the board)

(d) each hypothesis has a different set of bit values based on message information received from a MAC processor (first auxiliary request);

(e) decoding paths that correspond to decoded data bits that are inconsistent with the bit values specified by the hypothesis are removed from a set of possible decoding paths (second auxiliary request);

(f) the multiple hypotheses are generated in a hypothesis engine which uses predicted bit values by assuming different combinations of bit values (third auxiliary request);

(g) the multiple hypotheses are generated at a hypothesis engine and that the hypotheses are based on message type information received at the hypothesis engine from a MAC processor (fourth auxiliary request).

3.1 Inventive step (Article 56 EPC)

3.1.1 As to feature (d), D6 teaches that the set of hypothesised bit values relying on the a-priori scale information is based on the known or estimated physical bit parameters, namely the amplitude Ak of a received bit in position k and the associated standard deviation sigmak or variance sigmak**(2)(see e.g. paragraphs [0011] and [0051] to [0054]). Further, that information may evidently be delivered as "extrinsic information" that is indicative of adjustments in the confidence in the detected bit values based on earlier decoding processes in the system of D6 (see e.g. paragraphs [0007], [0009] and [0083], in conjunction with Fig. 8). In view of the fact that the processing for deriving such extrinsic information has to be performed at the medium access layer (MAC) in the underlying system (see in particular Fig. 8), the board finds that feature (d) is (at least) implicitly disclosed by D6.

3.1.2 Concerning feature (e), the board considers that the skilled person, confronted with the objective problem of "how to implement the selection of hypothesised candidate sequences in typical channel decoders", would be aware from his general knowledge that the proposed trellis-based (see e.g. "trellis states" in paragraph [0046] of D6) turbo decoding of convolutional codes according to D6 can equally be applied to the Viterbi decoding scheme (see e.g. paragraph [0092]). In a typical Viterbi decoder, disregarding trellis states and their corresponding decoding paths which are found to be inconsistent with the predetermined bits at the respective bit positions is known e.g. from D1 (see e.g. abstract; paragraph [0024], in conjunction with Figs. 4A and 4B and claim 1). Hence, given that the scaling factors do indeed correspond to bit values of the encoded data bits as claimed (cf. point 2.1.4 above) and contrary to the appellant's view, feature (e) cannot contribute to inventive step either.

3.1.3 Moreover, as to feature (f), the added use of a "hypothesis engine" refers simply to a certain labelling of the unit which is supposed to process the respective hypotheses, while using predicted bit values and assuming different combinations of those bit values corresponds directly to different sequences of scaling factors in the language of D6, contrary to the appellant's contention. Thus, feature (f) likewise cannot warrant an inventive step.

3.1.4 Lastly, as to feature (g), the board accepts that the indication of a message type as possible a-priori information is not derivable from D6 by any explicit statement or unambiguous implication. As to the question of obviousness, the board finds that it relates to the partial problem of "how to generate realistic hypotheses in the decoding system of D6" rather than "how to actually implement the selection of candidate sequences in the decoding system of D6" based on such hypotheses, underlying distinguishing feature (b') of claim 1 (see point 2.1.7 above). However, contrary to the appellant's opinion, D6 also teaches that estimated a-priori information such as Ak and sigmak**(2), which relate to different hypothesised sequences S, are in fact based e.g. on the "bit rate" used for the received bits (see e.g. paragraph [0054]). It is in turn commonly known that the bit rate used for the transmission of data messages may typically depend on the type of messages to be transmitted (such as control or data messages). Hence, the board holds that the skilled person would take up this hint to arrive at a solution according to feature (g), without exercising inventive skills. Thus, distinguishing features (b') and (g) represent a mere juxtaposition of obvious implementation measures.

3.2 Consequently, the first to fourth auxiliary requests are likewise not allowable under Article 56 EPC.

4. Request for remittal for further prosecution

4.1 The appellant requested that the case be remitted to the examining division for further prosecution in the light of newly introduced document D6 (cf. appellant's letter dated 31 May 2016, page 2, last paragraph).

4.2 The board, however, in the exercise of its discretion under Article 111(1) EPC, decides not to remit the case in view of the following observations:

- document D6, originating from the present appellant, was introduced by the board in reaction to fresh arguments and new claims according to the first auxiliary request submitted with the statement setting out the grounds of appeal (see also G 10/93, OJ EPO 1995, 172, Reasons 3, stating that, in ex parte proceedings, the Boards of Appeal are restricted neither to examination of the grounds for the contested decision nor to the facts and evidence on which the decision was based);

- the board, comprising technically qualified members, is in a position to judge the technical content of the claims and to deal with the merits of the present claim requests without remitting the case (see e.g. T 2249/13, Reasons 32).

5. Request for reimbursement of the appeal fee

5.1 The appellant further requested that the appeal fee be reimbursed on the grounds that the appeal regarding the decision and its objections was well-founded (cf. appellant's letter dated 31 May 2016, page 2, last paragraph).

According to Rule 103(1)(a) EPC, the appeal fee is reimbursed only if the board deems the appeal to be allowable, and reimbursement to be equitable by reason of a substantial procedural violation. However, the present appeal is not allowable (see points 2 and 3 above). Nor can the board see that the first-instance proceedings were tainted with any procedural violation, let alone a substantial one. Consequently, this request must be refused as well.