5G wireless networks harness the extensive spectrum available in the millimeter-wave (mmWave) frequency bands which set them apart from current wireless systems in terms of directivity, propagation loss, and susceptibility to blockages. Sub-6 GHz systems can attain omni-directional coverage, displaying limited sensitivity to physical obstacles. Still, they are incapable of achieving the same level of service quality as systems outfitted with electronically steerable directional antennas offering reduced propagation loss and higher gains due to the beam directionality. In our framework, we investigate the utilization of directional, steerable device-to-device (D2D) mmWave antennas as integral components. The nodes communicate by manipulating the orientation of their antennas, i.e., steering their beams. To minimize dependence on a base station, the D2D nodes are categorized into primary and secondary ones and they communicate in three phases: Uplink, Downlink, and PAPA. We delve into the impact of optimal steering of the main lobe beams transmitted by these D2D antennas as well as optimizing the time sharing among three phases. Within each phase, we assume that the nodes follow a random transmission scheduling scheme and derive the achievable rates accordingly. Through meticulous design of polynomial-time heuristics, we maximize the overall network capacity.