This is not a MATLAB question, so it isn't really appropriate here.
In my opinion, the answer is Yes, you need to discuss performance. It is common for improved security to come at the cost of increased cost.
Part of your thesis should be to analyze the big-O implementation cost for the base algorithm and for the improved algorithm. You would do that in order to show that the improved algorithm is somehow inherently slower, and by what rate. That would help satisfy the question of whether the problem is not the algorithm itself but rather your implementation skills -- could someone with different MATLAB experience potentially create significantly faster code? If the more secure algorithm were implemented in ASIC then how much slower would it be compared to the base algorithm.
How much you analyze might depend upon whether you are analyzing an existing security enhancement or if you are proposing your own enhancement. If it is your own enhancement then you should definitely include computation cost analysis. If you are analyzing something someone else has proposed and they already worked out costs then it is pointless for you to do that. If you are analyzing something someone else has proposed and they did not already work out costs, then cost analysis would be optional, but if the algorithm does turn out to be more secure than the base algorithm then cost analysis would be added value.
If the security conclusion is that the new algorithm is worse security (or no better) then cost does not matter. But if the general conclusion is that it is better security, then cost analysis would be a good idea. Part of the role of "promoting" something better is to "sell" people on it by showing that the improvement in functionality is worth the increased complexity. But if it turns out the new algorithm is only "a bit" better, not "significantly" better, then an outline of a cost analysis might be a good idea to demonstrate that the new algorithm is not "value for money"
