Since 1995, Machine to Machine (M2M) networks are spreading very quickly. These technologies allow both wired and wireless system communicating with other devices of the same type. Regardless of the type of machine or data, information usually is distributed in the same way, from a machine to a network, and then through a gateway to a system where it can be reviewed. This type of communications require new protocols and optimizations to minimize the energy consumption of the devices of the network to extent their battery life-times. The M2M network considered in this work is composed of a group of devices that are collecting data and, periodically, transmit this data to a coordinator upon request. Once each device is able to transmit the information correctly, passes into a sleep mode for saving energy. Generally, the use of random access protocols such as Aloha provides good results due to its low complexity. On the contrary, its performance is degraded drastically when the data traffic load increases or the number of devices is huge, which is the case in dense M2M networks. For that reason, new versions of the Aloha protocol were developed, such as Slotted-Aloha, which doubles the throughput of classical Aloha. Based on Slotted-Aloha, a new access protocol called Successive Interference Cancellation Frame Slotted Aloha (SICFSA) has been proposed in recent works. In SICFSA, the transmitter devices send multiple copies (replicas) of a data packet in different time slots. Each data packet indicates the time slots where the other replicas are transmitted. At the receiver side, i.e., the coordinator, it tries to recover the data from the transmitters by means of a Successive Interference Cancellation (SIC) algorithm. In order to successfully decode the transmitted packets, the coordinator has to identify the time-slots free of collisions. In this way, this information can be reused in the time-slots with collisions to decode data packets that have collided in the access to the channel. This work evaluates, by means of computer-based simulations using MATLAB and C, the delay and energy performance of a data-collection M2M network that uses SICFSA. The performance of SICFSA is compared with the one of conventional FSA. Results show that SICFSA outperforms FSA in about 50% in terms of delay and coordinator energy consumption, and in terms of devices energy consumption outperforms FSA in about 5%.