I have some sort of a misunderstanding regarding the Suzuki-Kasami distributed mutex algorithm. I am writing my question here because I failed on gaining access to the original paper.
I will call processes sites for convenience.
Why the token holder $j$ releases its token only when $RN_j[i]==LN[i]+1$ and not when $RN_j[i]>LN[j]$.
I mean, since we cannot assume anything on the quality of communication, maybe site $i$ had the time to increment his sequence number twice so that the token holder got a message with difference $2$ from $LN[j]$. In this case the token holder will not release his token and starvation will be applied.