Vehicular localization has witnessed significant attention due to the growing number of location-based services in vehicular cyber physical systems (VCPS). In vehicular localization, GPS outage is a challenging issue considering the growing urbanization including high rise buildings, multilevel flyovers and bridges. GPS-free and GPS-assisted cooperative localization techniques have been suggested in the literature for GPS outage. Due to the cost of infrastructure in GPS-free techniques, and the absence of location aware neighbors in cooperative techniques, efficient and scalable localization is a challenging task in VCPS. In this context, this paper proposes a geometry-based localization for GPS outage in VCPS (GeoLV). It is a GPS-assisted localization which reduces location-aware neighbor constraint of cooperative localization. GeoLV utilizes mathematical geometry to estimate vehicle location focusing on vehicular dynamics and road trajectory. The static and dynamic relocations are performed to reduce the impact of GPS outage on location-based services. A case study based comparative performance evaluation has been carried out to assess the efficiency and scalability of GeoLV. It is evident from the results that GeoLV handles both shorter and longer GPS outage problem better than the state-of-the-art techniques in VCPS.