Ask for some helps on DPC;

Now there are some question on DPC as follows:
1>on the classical DPC from Costa's description:
    i: background: from those descriptions of DPC in Costa's paper, when the interference is fully known to the transmitter, it can seem that the interference does not exist, in other words, the full capacity of point-to-point communication is not completely affected by this interference.  
    ii: implementation: take an example on its implementation for DPC:
         assume two independent data sources, e.g., symbol s1 and s2 are transmitted to one user,
        at transmitter, if s2 is pre-known to s1, for s1, we can use the DPC idea to do encoding and get, e.g., s1'. Then s1' + s2 is transmitted to this user.
        s1' and s2 share the same wireless channels.
        at receiver, we can use DPC decoding to obtain the original s1 and at the same time there is any interference from s2 due to DPC coding at transmitter.
        for the DPC coding and decoding, there are some typical algorithms, e.g., T-H pre-coding (modu) and others.
    iii: questions:
        a>is it right for my above understanding?
        b>if they are right, if we want to decode and obtain symbol s2 from the received signal expression, how to do it? In my opinion, s1' is the interference to s2 and there is capacity loss due to this interference for s2 transmission.
2>from the extension of DPC to, e.g., multi-user scenarios:
    i: background: usage of it under downlink MIMO broadcast channel (BC) for multi-user scenarios is a direct extension for classical DPC idea.
    ii: implemenatation: take an example on its implementation for DPC under this case:
        assume downlink MIMO BC, two users, u1 and u2. s1 and s2 are transmitted to two users.
        at transmitter, in order to do DPC, the following two conditions need to be satisfied:
            a>share transmission symbols, i.e., s1 and s2
            b>share two independent channel information, i.e., H1 for u1 and H2 for u2
        s1 and s2 do not share the same wireless channels
        at receivers, due to DPC coding at transmitter, receiver design at both users is simple, in other words, this system could be transformed into two parallel independent SISO channels.
    iii: questions:
        a>is it right for my above understanding?
        b>under multi-user scenarios, DPC could be just used under BC (point to multipoint) but not MAC (multipoint to point)? For the latter, SIC receiver may be a better choice especially from capacity view.
        
3>the differences between of both:
    i>If two data streams share the same wireless channel, the classical DPC idea may be appropriate but another stream could not be decoded without interference.
    under this case:
        pros: for DPC coding to one stream, just the knowledge of the other stream is needed and the shared channel does not needed at transmitter
        cons: one stream could be decoded without interference from the other stream but the performance of the other stream could not be satisfied if we want to decode it also
    ii>If two streams do not share the channel, i.e., transmitted to different users, the extension from DPC may be appropriate.

    under this case:
        pros: two streams could be well separated at users
        cons: two streams and all channels, e.g., H1 and H2 should be shared at transmitter
    iii: is my above summary right?

Don't pay too much time on DPC