In　wireless　sensor　networks(WSNs),　energy　efficiency　and　packet　time　delay　are　two　major　considerations　in　the　design　of　medium　access　control(MAC)　or　routing　algorithms.　However,　these　two　metrics　always　can＇t　be　satisfied　at　the　same　time,　and　existing　works　do　not　trade　off　them　well.　In　this　paper,　we　first　exploit　a　prediction　model　to　evaluate　the　amount　of　packet　in　the　next　period.　Then　we　formulate　energy　consumption　and　latency　as　a　multi-objective　optimization　function,　and　achieve　the　optimal　sleep　time　based　on　predicted　packet　number.　Finally,　we　propose　a　duty　cycle　adjusting　algorithm　for　WSNs.　Simulation　results　show　that　our　proposed　algorithm　can　improve　energy-efficiency　while　guaranteeing　packet　latency　is　minimized.