The optimal tracking problem is solved in the higher level controller where the interconnection of all the rooms and the walls are taken into consideration. The LQR controller minimizes a quadratic cost function which has two quadratic terms. One takes into account the comfort level and the other represents the control effort, i.e. the energy consumed to operate the HVAC system. There are two tuning parameters as the weight matrices for each of these two terms by which the performance of the controller can be tuned in different operating conditions. Simulation results show how much energy can be saved using this algorithm.