In another example, PDCCH may be not necessarily mapped to a single subframe/cell with the self-contained encoding scheme, so that an encoding scheme can be design without information bits distribution limits. For instance, PDCCH may be mapped with the combined encoding scheme. In this way, higher coding gain can be achieved.

For both examples, the UE may already start PDCCH decoding even if not all the related subframes of all related cells are received. In this way, lower latency can be achieved.

In order for the UE to know how to decode PDCCH generated as above, it has to know information about to which subframes the PDCCH is mapped to, to which serving cells the PDCCH is mapped to (in this case, the premise here is that the UE is active on that cell, e.g., due to carrier aggregation support), and/or the like. In an embodiment, the eNB may transmit the above information via signaling such as RRC, MAC CE, PHY signaling, so that the mapping can be static or dynamic. There may be ambiguity in this configuration signaling, e.g., the UE may know that 2 subframes/cells will would be used for joint encoding/decoding of PDCCH, yet not sure whether the current subframe/cell are the first or the second subframe/cell. In this case, it would be up to UE to do blind decoding, to test multiple hypotheses.

There may be different modes of PDCCH (with or without joint encoding/decoding as proposed above, or different numbers of subframes/cells where the joint encoding/decoding to be done) co-existing with each other, due to the different requirement of PDCCH reliability of different traffic types, i.e., C-MTC and non-C-MTC traffic, or different channel status of different UEs. There may be a variety of ways to differentiate various modes of PDCCH.