Considering　the　real-time　requirement　and　some　specific　limitations　(e.　g.　insufficient　computing　resource,　energy　constraint,　etc)　in　task　scheduling　of　wireless　sensor　networks,　different　priorities　were　assigned　to　tasks　according　to　their　deadline,　and　an　adaptive　task　allocation　algorithm　with　complicated　coalition　was　designed　through　analyzing　historical　information.　Moreover,　a　discrete　particle　swarm　optimization　algorithm　was　designed　via　employing　binary　matrix　coding　form.　The　proposed　optimization　algorithm　generates　coalitions　in　parallel　and　then　performs　subtask　allocation　algorithm　based　on　load　and　energy　balance.　Finally,　the　experimental　results　show　that　the　proposed　algorithm　strikes　a　good　balance　between　local　solution　and　global　exploration,　and　achieves　a　satisfactory　result　within　a　short　period　of　time.