Massive MIMO communication systems have been identified as one of the most prominent technologies of the next generation wireless standards, such as 5G, due to the large gains in energy and spectral efficiency that can be achieved. In the asymptotic condition of infinite number of antennas at the base station (BS), the performance bottleneck of these systems is due to the pilot contamination effect, i.e., the directional interference arising from users in adjacent cells that reuse the same set of orthogonal training sequences, and thus the interference seen by each user is determined by the pilot sequence assigned to him. We show in this paper that the system performance can be improved by appropriately assigning the pilot sequences to the users, in the so-called pilot allocation scheme. Depending on the optimization metric adopted, it is more advantageous to a user with certain long- term fading coefficient be assigned to a specific pilot sequence, whose interference can be completely estimated in advance by the BS by only knowing the long term fading coefficients of users in adjacent cells. Besides, if the objective is to maximize the number of users with a target quality of service, we have shown that the pilot allocation schemes can be combined with power control algorithms, resulting in much more improvements for the system. For unitary frequency reuse factor, we have found that the data throughput provided for 95% of the users increases when applying power control algorithm from 134kbps to 1.461Mbps with no pilot allocation, while this performance gain provided by power control changes from 793kbps to 6.743Mbps when pilot allocation is employed. If the reuse factor increases to 3, a 95%-likely data throughput of 17.310Mbps can be assured when pilot allocation and power control are suitably combined.