What do you mean by "sum rate of MUE n SUE separately"?

Usually, sum-rate of the overall network is calculated as sum-rate= datarates-of-MUEs + datarates-of-SUEs. In the case of individual tiers mostly, average data rates are plotted. 

I agree with your answer. But in the paper I have mentioned, authors have calculated sum rate for both the tires individually and then made a plot of sum rate of macro users versus the sum rate of small users. The number of macro users and small cell users obviously differs in each cell. 

This does not make much sense. Since both of the tiers have no specific relation (such as resource sharing etc),  in this case, I guess it does not provide any useful information to the readers.

Yes, I agree with that. Only possible reason(based on the suggestion by Emil Bjornson) I can think of is 'since the two tiers are sharing the system bandwidth, and if the total bandwidth is divided in different proportions between the macro user and small cell user. Then we can make a plot displaying the sum-rate of macro users versus small cell users, but how informative is it, can't figure that out. 

Only possible reason(based on the suggestion by Emil Bjornson) I can think of is 'since the two tiers are sharing the system bandwidth, and if the total bandwidth is divided in different proportions between the macro user and small cell user. Then we can make a plot displaying the sum-rate of macro users versus small cell users, but how informative is it, can't figure that out. 

The paper has only stated the total bandwidth, without mentioning about how it is going to be divided between macro and small cell users.

This is just my idea. I haven't done the simulation. 

Consider the power allocation. The small cell base stations, small cell and macrocell users have the same maximum power allocation. Considering Downlink or Uplink, the  power used for transmission by either group can be used for performance analysis .