This paper presents a new method for control of linear networked systems by combining the predictive control and the variable sampling period approaches. In this way, event-driven sensors are implemented, i.e. by construction the sensors are triggered to sample the outputs of the plant, when new control input signals are received by the actuators. In light of this formulation, at each sampling instant, total control loop delay will be equal to the sampling period which is unknown. In order to deal with the network effects associated with a range of pre-specified time delays, appropriate step invariant discrete-time models of the networked plant are calculated offline. Based on these, some stabilizing control signals are constructed online. The control signals are then packed in the control-side packet, transmitted back to the plant side and received by a time delay compensator module. A less conservative class of Lyapunov functions, called switched quadratic Lyapunov, is used here for stability analysis and stabilizing controller design. Simulation studies on well-known benchmark problems demonstrate the effectiveness of the proposed method.