Abstract-According to a recent European Union report, lighting represents a significant share of electricity costs and the goal of reducing lighting power consumption by 20% demands the coupling of light-emitting diode (LED) lights with smart sensors and communication networks. In this context, this paper proposes the integration of these three elements into a smart streetlight, incorporating a 24 GHz phased-array (Ph-A) front-end (FE). The main building blocks of this Ph-A FE integrated in a lowcost 90 nm complementary metal-oxide-semiconductor (CMOS) technology are fully characterized. The selected FE's architecture allows the implementation of transceivers as well as Doppler radar sensors functionalities. More specifically, the Ph-A technology is applied to a Doppler radar sensor in order to realize multi-lane road scanning and pedestrian detection. That way, the smart streetlight can become eco-friendly by turning on the LEDs only when necessary as well as to measure traffic parameters such as vehicle speed, type and direction. Intercommunication between the smart streetlights is based on a time-sharing mechanism that uses the same FE reconfigured as transceiver. Thanks to this functionality, the recorded traffic information can be relayed through adjacent streetlights to a control center, and control commands and warnings can be spread through the network. The system requirements are derived assuming a simplified model of the operating scenario with a typical inter-light distance of 50 m and line-of-sight between lights. The radar range is around 60 m, which allows for continuous coverage from one streetlight to the adjacent one. Meanwhile, a communication range of 140 m is derived as a fundamental requirement for reliable communication between streetlight sensors because it allows bypassing of one node in case of failure. For the developed building blocks -a low-noise amplifier, a variable-gain amplifier, a voltage-controlled oscillator and a vector modulation phase shifter -the design methodology is presented together with measurement results. The system power, consumption, noise figure and gain are estimated by means of a system analysis based on the measured data from the implemented blocks and the state of the art performances for the missing parts. It is shown that the requirements can be fulfilled with a total power consumption of around 375 mW in Doppler radar sensor mode and around 190 mW in transceiver mode. To the authors' knowledge, this kind of integration is new and overcomes some limitations of the currently used solutions based on infrared sensors and low-throughput communications.