Performance evaluation of an adaptive semi-persistent LTE packet scheduler for M2M communications

Large number of M2M devices are anticipated to be operating in future LTE networks which impose several system design challenges and wide range of service requirements. The LTE standard suffers from excess control channel overhead associated with radio resource allocation method for small, sporadic traffic per terminal which is often the nature of M2M communications. The rigid QoS support framework of LTE for limited number of voice and data services also fails to address the specific QoS requirements of M2M traffic classes. In this paper, we propose an adaptive LTE uplink scheduler which allocates radio resources to M2M traffic classes in either dynamic or adaptive semi-persistent manner based upon their traffic patterns and delay requirements. We also transform the concept of semi-persistent scheduling (SPS) implemented for VoIP scheduling in LTE to an adaptive SPS scheme which provides flexibility in allocated resource volume to accommodate changes in traffic dynamics. This new adaptation of SPS is particularly suitable for supporting random bursts of event-based M2M traffic yet has less control channel overhead than the dynamic scheduler. We demonstrate from simulation results that the proposed adaptive scheduler can maximize uplink data capacity by reducing dependency on downlink control channel as well as satisfy the QoS requirements of different M2M traffic classes compared to full dynamic and rigid SPS approaches.