T 2083/09 (Echo cancellation/LUCENT) 30-01-2012

Reasons for the decision

1. Claim 1 - inventive step (Article 56 EPC)

1.1 The invention generally relates to echo cancellation in data communication between modems over a communication network. The invention specifically aims at improving echo cancellation when the analogue local loop signal output by the modem is converted into a digital signal and further companded for the purpose of transmission over the communication network.

1.2 D3 is in the same technical field as the application and refers to the problem of properly training an echo canceller using inter alia a half-duplex approach when the transmitted signal is additionally subjected to a non-linear compression / expansion operation (cf. column 1, lines 37 to 58). D3 is therefore considered by the board as the single most relevant prior art document for assessing inventive step.

Each modem 100 and 300 in D3 is separately configured to cancel echo in the transmitted signal (column 5, line 30 to column 6, line 7). An echo canceller 650 (figure 6) as provided in modem 100 is operated in time sequence either as a fixed echo canceller, for cancelling echo during data exchange (step 735 in figure 3) or as an adaptive echo canceller, for determining the filter coefficients for the subsequent data exchange phase during a training phase (step 725 in figure 3). The fact that the echo cancellers are solely trained during the training phase, which is before data exchange, is understood by the board as meaning that the filter coefficients of the echo canceller are not modified during the data call.

1.3 Accordingly, the echo canceller of claim 1 differs from that of D3 as regards the following features:

(a) providing a separate adaptive echo canceller, a comparator for determining whether the adaptive echo canceller is providing increased echo cancellation in comparison to the fixed echo canceller and a memory element for storing the tap coefficient values detected by the adaptive echo canceller; and

(b) providing an echo-cancelled signal before performing a companding operation on the current data call.

1.4 In the board's view the claim defines an aggregation or collocation of features, in which the technical problem when starting out from D3 as the most relevant prior art is composed of two separate, mutually independent sub-problems: the first sub-problem, addressed by feature (a), is to provide a concrete configuration for the echo canceller and the second sub-problem, addressed by feature (b), consists in having an appropriate location for the echo canceller in the signal path. According to the established jurisprudence of the boards of appeal the contribution to inventive step of features (a) and (b) has therefore to be considered separately (cf. T 130/89, OJ 1991, 514).

1.5 Regarding feature (a), D2 discloses a configuration of an echo canceller (figure 4) including a programmable filter 22, an adaptive filter 20 and a transfer control logic 23. The programmable filter performs echo cancelling on an outgoing signal, the adaptive filter determines filter coefficients for updating the coefficients of the programmable filter (column 5, lines 17-22) and the transfer control logic compares the performance of echo cancellation achieved by the adaptive filter with that achieved by the programmable filter. The D2 echo canceller implicitly contains a memory for storing the filter coefficients of the programmable filter since it is said that the filter coefficients are not allowed to be updated during periods of doubletalk (column 5 lines 23 to 30). Thus, in operation the programmable filter uses coefficients which were earlier determined by the adaptive filter. It is noted that the echo canceller of D2 does not mention whether the coefficients of the fixed echo canceller can be updated during a call if the adaptive echo canceller provides better echo cancellation. In the board's view it is however a matter of non-inventive design choice for the skilled person whether the filter coefficients are allowed to be updated during a call or whether they are only updated after the call for use during the subsequent call. The skilled person, seeking for an appropriate configuration of the echo canceller of D3, would be led by D2 to a configuration according to feature (a).

Regarding feature (b), D6 discloses a telephony system in which echo canceller 116 is provided at a location in the signal path at which the received and transmitted signals Rx, Tx are present in a linear version (cf. figure 11 and page 8 line 30 to page 9 line 12 of D6). Starting out from D3, the skilled person would therefore be led by D6 to a configuration of the echo canceller such that an echo-cancelled signal is provided before performing a companding operation on the current data call.

1.6 In the letter received on 3 November 2011 the appellant took the view that the echo canceller according to claim 1 was configured for processing a digital signal whereas the echo cancellers disclosed in D3 and D6 were not.

The board notes that echo canceller 655 in D3 is positioned at the signal path downstream of A/D-converter 625 from which the board concludes that the signal at the echo canceller is present in a digital version. Similarly, the signal at the echo canceller 116 in figure 11 of D6 is a signal within digital signal processor DSP1 and is therefore implicitly in a digital version.

The appellant's argument that echo cancellation would not be carried out on a digital signal in either of D3 or D6 must therefore fail.

1.7 In conclusion, the skilled person, starting out from D3 and having regard to D2 and D6 would be led to the echo canceller of claim 1 without the exercise of inventive skill (Article 56 EPC).

2. There is accordingly no request on file on the basis of which the appeal could be allowed. The appeal must therefore be dismissed.