Users face privacy risks when providing personal location data to untrusted location based services (LBSs). Many existing attempts to anonymize users have relied on suppressing information about users' locations. This suppression results in reduced functionality, including degraded spatial accuracy or the loss of realtime location updates. We propose anonymizing a user by camouflaging their current location with various predicted paths. This is akin to increasing the noise floor around the user. We believe this different approach provides many advantages. By allowing the user to keep all desired functionality while still protecting the user's privacy, we enable use of a broader class of LBSs. To demonstrate our proposal, we develop CacheCloak, a system that enables realtime anonymization of continuous location updates without any loss of accuracy. Data from a GIS-based traffic simulation in an urban environment shows that CacheCloak can anonymize users rapidly.