Intrusion detection is one of the most important services in a smart home, which requires to monitor intrusion events and to react against them. A Wireless Sensor and Actor Network (WSAN) has a set of sensor nodes for monitoring events and a set of high capability nodes, called actor nodes, for reacting to the events. It can provide an infrastructure for building the intrusion detection system in a smart home. We need to jointly consider fault-tolerance and timeliness issues because sensor nodes are likely to be influenced by failures (e.g., dead battery) and need to deliver packets in real time over wireless links vulnerable to interference caused by walls, floors or furniture. In this paper, we propose Optimal Multipath Planning (or OMP), which is based on Edmonds- Karp maximum flow algorithm and Goldberg and Tarjan minimum cost flow algorithm, for WSANs to set up paths to deliver intrusion event notification from event sources to actors in a smart home. The planning is optimal in the sense that it sets up the maximum number of node-disjoint paths of links with the minimized expected transmission count (ETX). We also evaluate OMP's performance by simulations and compare it with the minimum cost planning using Dijkstra algorithm to show its advantages.