In recent decades, traffic congestion has become a major issue in traffic networks, especially in urban networks comprised of a set of short links and signalized intersections. To circumvent the issue, various traffic control and management strategies have been devised; however, the proposed strategies are rarely developed at network-wide level. Further, the modelling approach is based on microscopic models that cannot be adopted for centralized control or macroscopic models that have limited capacity to properly describe important phenomenon of traffic networks. This thesis aims at modelling and control of a large-scale urban network comprised of multiple pockets of congestion. The modelling approach is based on macroscopic fundamental diagram (MFD), which assumes a well-defined relationship between average flow and average density for any traffic network with spatially homogeneous distribution of vehicles. This simpler representation of large-scale traffic networks using aggregated traffic variables facilitates a centralized and real-time control of urban networks. To use the system-wide benefits of MFD models, firstly, an anticipatory control scheme, integrating road users routing responses to the control model is advanced. The proposed anticipatory control approach is found to produce globally optimal solutions and move the network towards system optimum traffic condition. Thereafter, a proportionally fair control scheme that simultaneously enhances efficiency and fairness among road users is developed. The unique feature of the developed perimeter controller is consideration of road users' trip utility in the control model without much sacrificing efficiency for fairness. Despite the computational advantages of aggregated MFD models, these models do not describe important phenomenon of traffic networks. In the final part of this thesis, the MFD dynamics is enhanced to capture multiple kinematic waves, congestion, and queueing with high precision and within reasonable computational effort. Further, an approach for incorporating connected/autonomous vehicles (CAV)s into MFD dynamics is introduced; the network-wide effect of CAVs on network's traffic state is then investigated.