Multi-goal pathfinding (MGPF) is a problem of searching for a path between a start and a destination allowing a set of goals to be satisfied. We address MGPF in ubiquitous environments that accommodate cyber, physical and social (CPS) entities from smart objects to sensors and to humans. Given a MGPF problem in a pervasive environment, our approach aims at exploiting data from various resources including CPS entities located in the environment and external resources such as the Web to solve the problem. In this paper, we present a knowledge model for describing a ubiquitous environment integrating its spatial dimension, CPS entities it contains and its relevant resources. A global view of the approach is provided. We address particularly one of the challenges in MGPF, namely goal satisfaction problem, which consists of identifying through which entities a goal can be satisfied. Towards this aim, we design an ontology to formally model CPS entities, goals and their relations. We describe a method to exploit modeled knowledge in order to solve the goal satisfaction problem.