In 5G cellular networks Machine type communication devices (MTCD) for Machine-to-Machine (M2M) communication are subjected to the problem of collisions, delay, scalability and quality of service (QoS) during data transmission. The M2M communication has become ubiquitous into Internet of things (IoT). A power efficient MAC switching protocol and adaptive Time Division Multiple Access (PMAC-ATDMA) is presented to solve the issue of collision, QoS and scalability in this paper for M2M Communication. This consists of grouping, dynamic MAC switching and allotment of timeslots. In grouping processing of MTCDs, using Harris Hawks Optimization (HHO) algorithm is used for selecting a head i.e. MTC head (MTCH). CSMA/CA and CSMA/CARP protocols differ in back off time are used, switching between them happens based on the density, backlogged devices and active nodes that reduces collisions. The timeslots are assigned by ATDMA, in accordance with the data size and requirements of QoS. The problem of synchronization in traditional TDMA is overcome by the use of Markov chain model the simulation of this PMAC-ATDMA is performed in network simulator tool. The evaluation is performed in terms of access delay, energy, probability of collision and successful packet transmissions.In 5G cellular networks Machine type communication devices (MTCD) for Machine-to-Machine (M2M) communication are subjected to the problem of collisions, delay, scalability and quality of service (QoS) during data transmission. The M2M communication has become ubiquitous into Internet of things (IoT). A power efficient MAC switching protocol and adaptive Time Division Multiple Access (PMAC-ATDMA) is presented to solve the issue of collision, QoS and scalability in this paper for M2M Communication. This consists of grouping, dynamic MAC switching and allotment of timeslots. In grouping processing of MTCDs, using Harris Hawks Optimization (HHO) algorithm is used for selecting a head i.e. MTC head (MTCH). CSMA/CA and CSMA/CARP protocols differ in back off time are used, switching between them happens based on the density, backlogged devices and active nodes that reduces collisions. The timeslots are assigned by ATDMA, in accordance with the data size and requirements of QoS. The problem of synchronization in traditional TDMA is overcome by the use of Markov chain model the simulation of this PMAC-ATDMA is performed in network simulator tool. The evaluation is performed in terms of access delay, energy, probability of collision and successful packet transmissions.