Accurate wireless traffic forecasting facilitates proactive adaptation in wearable-based localization systems, enhancing reliability under dynamic conditions. We propose fully-connected echo state network (FCESN), a lightweight prediction model to strengthen wearable localization performance by anticipating communication trends in enterprise networks. FCESN integrates reservoir computing with a fully-connected layer, achieving high accuracy and efficiency in predictions. Evaluated on real-world data from specific access points for day-ahead forecasting, our proposal surpasses state-of-the-art models in precision and trend preservation while demanding significantly lower computation. The predicted traffic patterns are leveraged to proactively adjust localization strategies, enabling robust, context-aware positioning in dynamic enterprise environments.