Interest in exchange bias (EB) in magnetic nanoparticles has increased in the past few years by virtue of its potential for application in fields such as ultrahigh-density magnetic recording. A modified Monte Carlo Metropolis method is performed to simulate the effect of cooling field on EB and asymmetric reversal modes of a granular system of ferromagnetic (FM) nanoparticles embedded in an antiferromagnetic (AFM) matrix, based on three-dimensional classical Heisenberg model. The results show that the EB first decreases slightly due to the energy barriers in the antiferromagnet, while the coercivity and vertical magnetization shift increase with the increase of cooling field, finally, they all level off as the cooling field is strong enough. Whereas the cooling- and measuring-field angular dependence of asymmetric reversal modes reveal asymmetric reversal mechanism and interesting rotation process of FM spins, confirming the existence of the net magnetization on the surface of AFM matrix. The reason may be due to the energy competition and geometric frustration of system. However, the strong interfacial coupling may change the intrinsic atomic configuration of antiferromagnet to influence the EB and reversal modes.