A method includes receiving first and second synchronization sequences that jointly indicate an identity of at least one base station. Multiple hypotheses are selected, each hypothesis associating a respective first candidate sequence with the first synchronization sequence and a respective second candidate sequence with the second synchronization sequence. Each hypothesis is assigned a respective composite score depending on a respective first quality of match between the first candidate sequence in the hypothesis and the first synchronization sequence, on a respective second quality of match between the second candidate sequence in the hypothesis and the second synchronization sequence, and on a similarity between the respective first and second qualities of match. The at least one base station is identified based on respective composite scores of the hypotheses.