Data Providers vs Players

[Thanos Giannopoulos]

Hello,

I am experimenting with the SCALE-MAMBA software and I was wondering if it is possible to have the data providers separate from the players.
For example, imagine a protocol that runs by 3 players but the input data are coming from 5 different providers.

Is something like that possible, and if so what is the procedure that these 5 providers should follow in order to secret-share their data and feed them into the protocol so the 3 players can make computations on them?

I couldn't find something similar in the provided examples.

Thanks a lot in advance,
Thanos

[Nigel Smart]

Look at the IO class, you will need to extend this
in some way. An issue is that data providers CANNOT
prepare sharings exactly as in SPDZ due to the MACs.

An obvious fix would be for the data provider to
split x into 3 things, x1,x2,x3 additively.
Then send xi to player Pi, who then uses private_input
to enter the xi. Then the shared [xi] values
are added up to obtain [x].

Nigel

> --
> You received this message because you are subscribed to the Google
> Groups "SPDZ/SCALE-MAMBA Discussion Group" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to spdz+uns...@googlegroups.com
> <mailto:spdz+uns...@googlegroups.com>.
> To post to this group, send email to sp...@googlegroups.com
> <mailto:sp...@googlegroups.com>.
> Visit this group at https://groups.google.com/group/spdz.
> To view this discussion on the web, visit
> https://groups.google.com/d/msgid/spdz/b365dcc5-86c7-430e-b319-80cf91d043fd%40googlegroups.com
> <https://groups.google.com/d/msgid/spdz/b365dcc5-86c7-430e-b319-80cf91d043fd%40googlegroups.com?utm_medium=email&utm_source=footer>.
> For more options, visit https://groups.google.com/d/optout.

[Thanos Giannopoulos]

Thanks for the reply,

So if the shares are not MAC'd how is it possible for SCALE-MAMBA to perform the SMPC protocol on them?
Aren't the MAC checks necessary?

Also, if I understand you correctly, that is a procedure that I would have to implement and is not provided with the current SCALE-MAMBA package right?

Thanks again,
Thanos

[Nigel Smart]

On 03/09/2018 14:19, Thanos Giannopoulos wrote:
> Thanks for the reply,
>
> So if the shares are not MAC'd how is it possible for SCALE-MAMBA to
> perform the SMPC protocol on them?
> Aren't the MAC checks necessary?

These are added when you do private_input to each of the players.
The five players just do a naive 3-out-of-3 sharing of their data

>
> Also, if I understand you correctly, that is a procedure that I would
> have to implement and is not provided with the current SCALE-MAMBA
> package right?
>

Indeed. Because you are building an 8 party protocol using SCALE
for 3 of the parties. So you need to build the protocol for
the 5 to talk to the three yourself. But that is just TLS plus
a bit of calling the SCALE C-code yourself

Nigel

[Dragos Rotaru]

You need to be careful as Nigel's method works if the SCALE parties are honest when they input the shares provided by th external parties.

If you want your protocol to be resilient to SCALE with malicious parties you should check out the Input Supply (Fig 2) from this paper: https://eprint.iacr.org/2015/1006

Dragoș

To unsubscribe from this group and stop receiving emails from it, send an email to spdz+uns...@googlegroups.com.
To post to this group, send email to sp...@googlegroups.com.

Visit this group at https://groups.google.com/group/spdz.

To view this discussion on the web, visit https://groups.google.com/d/msgid/spdz/af970bb5-82eb-403e-a0a2-9ab4d10b5648%40googlegroups.com.