In large-scale wireless systems, networks are highly dynamic and nodes are not fully reliable due to limited energy resources or failures. Wireless multipath fading links make the network data delivery more challenging. To cope with network dynamics and link failures, we propose a novel network modulation (NeMo) approach for OFDM over large-scale wireless systems. NeMo enhances data persistence by guaranteeing the symbol recovery at the sink(s) without pre-storing or delaying the decoding process by collecting enough symbols. We consider practical channel conditions and show that NeMo enjoys diversity from grouping and subcarrier mapping strategies in OFDM. The simulation results confirm that NeMo becomes more efficient when it is combined with OFDM in terms of reliability and successful data persistence for unreliable wireless networks.