Crypto Protocol Evaluation - Is this algorithm susceptible to a known plaintext attack?

Suppose we have two parties, A and B.
A sends a nonce, n1, to B.
B sends back a nonce, n2, and an authentication payload (hash), auth.

The authentication payload, auth, is:
pseudo-random-function (128-bit key, n1|n2)

The nonces ofcourse are random for each session.

The pseudo-random-function is a subset of SHA, usually SHA-256.

If the adversary has access to everything but the 128-bit key, is this protocol secure against known plaintext attacks?
Any help would be greatly appreciated.

Hi,

Is this a function you're deriving yourself, or a known function?

Is it supposed to be one-way or reversible? Your mentioning of SHA suggests one-way...

Best regards,
AstroTux.

Hi Astro Tux,

The pseudo-random function is a known function, most likely to be SHA-256. And yes, it will be one-way. I hope this clarifies the problem statement.

I'm not seeing any value in the two nonces. You are passing them in the clear, as you say, you assume that Mallet has them. So the strength is just that of the key and your algorithm. The nonce values are just noise.

Thanks for the clarification.

I don't quite understand the aim of this though...?

Best regards,
AstroTux.