While horseshoe filterless networks with point-to- multipoint (P2MP) transceivers offer a cost-effective solution with inherent survivability against single-link failures, certain scenarios may demand an even higher degree of network resilience. This is particularly true in regions with frequent power outages and/or where load shedding is enforced. Addressing this problem requires a careful design and trade-off between survivability and cost . This paper describes a multi-objective optimization framework based on Integer Linear Programming (ILP) that accurately models the problem and can be used to obtain solutions that prioritize one or the other metric. Moreover, the paper reports a study using a set of randomly generated horseshoe topologies that demonstrates how strategically placing amplifiers and unbalanced couplers can significantly enhance network resilience, while also keeping the minimum number of additional optical amplifiers required.