Agame-theoretical approach to network capacity planning under competition

The paper discusses the dimensioning strategies of two network providers (operators) that supply channels to the same population of users in a competitive environment. Usersare assumed to compete for best service (lowest blocking probability of new request), while operators wishto maximize their profits. This setting gives rise to two interconnected, noncooperative games: a) a users game, in which the partition of primary traffic between operators is determined by the operators' channel capacities and by the users' blocking-avoidance strategy; and b) a network dimensioning game between operators in which the players alternate dimensioning decisions thatmaximize their profit rate under the current channel capacity of his/her opponent. At least for two plausible users' blocking avoidance strategies discussed in the paper, the users game will always reach some algorithmic equilibrium. In the operators' game, the player strategies are given by their numbers of deployed chanels, limited by their available infrastructure resources. If the infrastrucutre is under-dimensioned with respect to the traffic rate, the operators game willreach a Nash equilibrium when both players reach full use of their available infrastructures. Otherweise, a Nash equilibrium may also arise if both operators incur the same deployment costs. If costs are asymmetric, though, the alternating game may enter a loop. If the asymmetry is modest, both players may then try to achieve a competitive monopoly in which the opponent is forced to leave the game or operate with a loss (negative profit). However, if the asymmetry is high enough, only the player with the lower costs can force his opponent to leave the game while still holding a profitable operation.