Vehicular　networks　have　become　as　an　important　platform　to　monitor　metropolitan-scale　traffic　information.　However,　it　is　a　challenge　to　deliver　and　process　the　huge　amount　of　data　from　vehicular　devices　to　a　data　center.　By　studying　a　large　number　of　taxi　data　collected　from　around　3,000　taxis　from　Shenzhen　city　in　China,　we　find　that　the　data　readings　collected　by　vehicular　devices　have　a　strong　spatial　correlation.　In　this　paper,　we　propose　a　novel　scheme　based　on　compressive　sensing　for　traffic　monitoring　in　vehicular　networks.　In　this　scheme,　we　construct　a　new　type　of　random　matrix　with　only　one　nonzero　element　of　each　row,　which　can　significantly　reduce　the　number　of　data　needed　to　be　transmitted　while　guaranteeing　good　reconstruction　quality　at　the　data　center.　Simulation　results　demonstrate　that　our　scheme　can　achieve　high　reconstruction　accuracy　at　a　much　lower　sampling　rate.