The process of computing routes that network traffic must follow throughout network has become much more complex
in recent years. PCE (Path Computational Element) technology is emerging and gaining importance under the
circumstances. In this paper, PCE architecture is outlined, and the impact of PCEs allocation decisions is discussed
briefly. To track the problem of locating PCEs, an integer linear programming (ILP) model is presented to find the
optimal PCEs allocation solution in multi-domain optical networks. The objective is to minimize average amount of time
for sending a message to all nodes in the topology, i.e. message flooding cost. Then, two heuristics, LSPLP and TSPLP,
are developed based on this model. Numerical results show that compared with traditional allocation strategies, the
proposed algorithms can reduce the message flooding cost efficiently.