Multi-user orthogonal frequency division multiplexing(OFDM) is one of the most important techniques for next-generation wireless communication systems.The sum capacity and proportional fairness index are two important metrics in OFDM adaptive resource allocation.In existing researches,the two metrics have individually yielded approximately optimal solutions.However,as the two goals seem to contradict each other,it means that they cannot achieve optima simultaneously.This paper presents a resource allocation algorithm based on subcarrier exchange.By relaxing the fairness index,the proposed algorithm can dynamically adjust the fairness requirements.Furthermore,By considering a criterion that minimizes the loss of sum capacity,the proposed algorithm can balance the tradeoff between fairness and sum capacity.In contrast to existing algorithms,our proposal improves performance by approximately 30% in terms of sum capacity,at the expense of approximately 10% reduction in proportional fairness.Moreover,if a malicious user exists,each user in the system can obtain higher ergodic capacity compared with the conventional algorithms.To decrease the complexity,a low-complexity simplified alternative is presented,which outperforms the conventional algorithm with an increase in the number of users.Four numerical simulations are carried out to verify the efficiency and effectiveness.Finally,we discuss in detail the computational complexity of the proposed algorithm. Multi-user orthogonal frequency division multiplexing (OFDM) is one of the most important techniques for next-generation wireless communication systems. The sum capacity and proportional fairness index are two important metrics in OFDM adaptive resource allocation. In existing researches, the two metrics have Individually yielded approximately optimal solutions.However, as the two goals seem to contradict each other, it means that they can not achieve optima .This paper presents a resource allocation algorithm based on subcarrier exchange. adjust the fairness requirements.Furthermore, By considering a criterion that minimizes the loss of sum capacity, the proposed algorithm can balance the tradeoff between fairness and sum capacity. In contrast to existing algorithms, our proposal improves performance by approximately 30% in terms of sum capacity, at the expense of approximately 10% reduction in proportional fairnes s.Moreover, if a malicious user exists, each user in the system can obtain higher ergodic capacity compared with the conventional algorithms.To decrease the complexity, a low-complexity simplified alternative is presented, which outperforms the conventional algorithm with an increase in the number of users.Four numerical simulations are carried out to verify the efficiency and effectiveness. Finally, we discuss in detail the computational complexity of the proposed algorithm.