Due　to　the　uncertainty　of　elements　in　the　random　matrix,　the　design　of　composite　probes　on　compressed　sensing　microarray　(CSM)　becomes　more　complexity.　In　this　paper,　we　proposed　a　sparse　random　measurement　matrix　with　`　0/　1＇　binary　element,　and　fixed　the　same　amount　of　elements　`　1＇　on　each　row,　to　construct　the　CSM　composite　probe.　There　is　the　same　dilution　for　the　mixed　solution　of　target　segments　to　ensure　the　consistency　of　gene　concentration,　so　the　composite　probes　which　made　up　of　the　linear　combination　of　target　segments　are　very　simple.　Simulation　experiment　results　show　that　the　variation　characteristics　of　the　target　segment　can　be　accurately　recovered　by　OMP　algorithm　under　N　=　96　sequence　segments　and　variation　sparsity　level　K　＜=　12,　when　M　=　48　composite　probes　are　constructed　with　a　sparse　random　matrix　fixed　amount　of　non-zero　elements　each　row.