The　emerging　5G　wireless　networks　will　pose　extreme　requirements　such　as　high　throughput　and　low　latency.　Caching　as　a　promising　technology　can　effectively　decrease　latency　and　provide　customized　services　based　on　group　users　behaviour　(GUB).　In　this　paper,　we　carry　out　the　energy　efficiency　analysis　in　the　cache-enabled　hyper　cellular　networks　(HCNs),　where　the　macro　cells　and　small　cells　(SCs)　are　deployed　heterogeneously　with　the　control　and　user　plane　(C/U)　split.　Benefiting　from　the　assistance　of　macro　cells,　a　novel　access　scheme　is　proposed　according　to　both　user　interest　and　fairness　of　service,　where　the　SCs　can　turn　into　semi-　sleep　mode.　Expressions　of　coverage　probability,　throughput　and　energy　efficiency　(EE)　are　derived　analytically　as　the　functions　of　key　parameters,　including　the　cache　ability,　search　radius　and　backhaul　limitation.　Numerical　results　show　that　the　proposed　scheme　in　HCNs　can　increase　the　network　coverage　probability　by　more　than　200%　compared　with　the　single-　tier　networks.　The　network　EE　can　be　improved　by　54%　than　the　nearest　access　scheme,　with　larger　research　radius　and　higher　SC　cache　capacity　under　lower　traffic　load.　Our　performance　study　provides　insights　into　the　efficient　use　of　cache　in　the　5G　software　defined　networking　(SDN).　©　2016　IEEE.