Abstract: In wireless sensor networks, the contention-based S-MAC protocol has the problem that the node's activity time cannot be dynamically changed according to the communication load and the delay caused by node sleep. This paper combines the T-MAC protocol and the D-MAC protocol and proposes an improved method to address these problems. Simulation results show that the proposed improved method can not only adapt the node's activity time to the dynamic changes of the communication load, but also greatly reduce the delay caused by node sleep. Keywords: S-MAC; communication load; delay; NS-2 1 Introduction With the development of computers, sensors and wireless communication, a new type of computer network has emerged - wireless sensor network [1]. Due to its wide application prospects, wireless sensor network has been favored by more and more researchers. The Media Access Control (MAC) protocol determines the way wireless channels are used, allocates limited communication resources between sensor nodes, and uses them to build the underlying infrastructure of the sensor network system. The MAC protocol has a great influence on the performance of sensor networks and is one of the key network protocols to ensure efficient communication in wireless sensor networks. S-MAC [2] protocol is a typical contention-based random access MAC protocol. It is a sensor network MAC protocol proposed based on the IEEE 802.11 MAC [3] protocol to meet the energy-saving requirements of sensor networks. This protocol has good scalability and does not require strict time synchronization, but it also has the problems that the node activity time cannot be dynamically changed according to the communication load and the delay caused by node sleep. This paper first analyzes the mechanism adopted by the S-MAC protocol and points out its problems. Then, combined with the T-MAC protocol and D-MAC protocol, it proposes an improvement method. Finally, through simulation analysis, it is proved that the improved method can make the node activity time flexibly adapt to the changes in network communication load, further save energy, and at the same time, greatly reduce the delay caused by node sleep. 2 Analysis of the mechanism adopted by the S-MAC protocol The S-MAC protocol is a contention-based control protocol. It mainly adopts the following mechanism: "Virtual cluster" mechanism: Each node broadcasts a synchronization data packet containing its own scheduling information at the beginning of each time slot. The node that receives the synchronization data packet adjusts its clock as needed. In this way, nodes with the same scheduling form a "virtual cluster". In principle, the entire network should operate on the same "time-slot structure," but due to mobility and time-slot scheduling mechanisms, the network can contain many "virtual clusters." Periodic activity and sleep mechanisms: The S-MAC protocol divides time into multiple frames, each consisting of two parts: an active state and a sleep state. In the active state, nodes communicate with neighboring nodes, receiving or sending data; the active state is typically fixed at 300ms. In the sleep state, nodes turn off their transmitters and receivers to reduce energy consumption. If data needs processing at this time, it is buffered and processed when the node becomes active again. This periodic activity/sleep significantly reduces energy consumption caused by idle listening. 3. Problems with the S-MAC Protocol: The S-MAC protocol uses a periodic activity/sleep scheduling mechanism. The active time is usually fixed, while the message rate varies. The length of time the protocol is in the active state cannot be dynamically adjusted according to changes in network traffic, thus failing to effectively save energy. Nodes whose communication modules are in a dormant state must wait until the communication module switches to an active cycle before sending data if an event is detected. When an intermediate node needs to forward data, the next-hop node may be in a dormant state, and it must also wait for the next node to switch to an active cycle. This latency caused by node dormancy increases proportionally with the number of hops in the path. 4. Improvements to the S-MAC Protocol To address the issues of the S-MAC protocol in wireless sensor networks, such as its inability to dynamically adjust node activity time according to communication load and the latency caused by node dormancy, we first analyze the improvements made by the T-MAC and D-MAC protocols, and then propose our own improvement method. The T-MAC protocol was mainly proposed to address the problem of the fixed activity time of the S-MAC protocol, which cannot adapt to changes in communication load. It dynamically adjusts the activity time according to communication traffic while maintaining a constant cycle length. Nodes in an active state, if no cycle timer overflow or sensing network conflict occurs within a given time period (TA), end their active state and enter a dormant state, but this introduces the problem of early dormancy. The D-MAC protocol was mainly proposed to address the latency problem caused by node dormancy. It employs an interleaved scheduling mechanism, dividing the node cycle into receive time, send time, and sleep time. The receive time and send time are equal, each representing the time required to send a data packet. Each node's scheduling has a different offset, with the send time of a lower-level node corresponding to the receive time of an upper-level node. This allows data to be continuously transmitted from the data source node to the aggregation node, reducing transmission latency. However, the D-MAC protocol requires strict time synchronization. Combining the advantages and disadvantages of both T-MAC and D-MAC protocols, an improved method for the S-MAC protocol is proposed below. The improved S-MAC protocol mechanism is shown in Figure 1: [align=center] Figure 1 S-MAC Improvement[/align] The node cycle is still divided into active and sleep times, neither of which are fixed. In the improved protocol, nodes periodically wake up to listen. If no activation event occurs within a time interval Δt, the active cycle ends early, reducing energy consumption. Activation events include: detecting a collision on the network; starting the periodic frame timer; ending data transmission and waiting for confirmation from the other party; listening for RTS and CTS packets on the network and exchanging data with neighboring nodes. Each node's scheduling has a different offset, with the activity time of lower-level nodes corresponding to that of upper-level nodes. This ensures that when a lower-level node needs to send data to or receive data from a destination node, it can quickly and promptly communicate with its upper-level neighbors, reaching the destination node much like climbing stairs or descending a flight of stairs. This significantly reduces latency issues caused by node sleep. The improved S-MAC protocol employs an ACK acknowledgment mechanism; if a sending node does not receive an ACK acknowledgment, it must retransmit at the next transmission time. Once a node correctly receives data, it immediately sends an ACK message to the node that sent the data. To reduce data transmission conflicts, nodes wait a fixed backoff time and then randomly select a transmission waiting time within the conflict window. Furthermore, an adaptive duty cycle mechanism is used to dynamically adjust the activity time of nodes along the entire path based on network traffic changes. When the amount of data transmitted is large, a node may need to send multiple packets within a transmission cycle. In this case, the node's duty cycle needs to be increased, and nodes along the transmission path are requested to increase their duty cycles accordingly. By performing hop-by-hop reservations along the transmission path, the network's data transmission rate can be greatly improved. 5 Simulation Analysis This paper uses NS-2, an object-oriented, discrete event-driven network environment simulator developed by UC Berkeley, to simulate the improved S-MAC protocol. TOSSIM can also be used as a wireless sensor network simulation environment [4]. The simulation software NS-2 takes a script as input, which describes the network topology, network protocol, network load, and some control parameters. NS-2 outputs a series of data, such as the number of data packets sent by each data source, the delay at each network node, etc. In the simulation experiment, we collected the following performance evaluation parameters for the original S-MAC protocol and the improved S-MAC protocol to compare and analyze them: energy overhead and delay. The activity time is judged based on the amount of energy consumed. Their definitions are as follows: Energy overhead: The total cost of sending a certain number of packets from the source point to the destination node. Delay: The end-to-end delay of transmitting a packet. The simulation parameters are set as follows: wireless device bandwidth 100kbps, transmission range 250m, interference range 550m, packet length 100 bytes, transmission power 0.66 watts, reception power 0.395 watts, power consumption during idle listening 0.35 watts, and power consumption during sleep is negligible and set to 0. Based on the wireless parameters and packet length, the S-MAC protocol activity time is set to 20ms. The activity time of the improved S-MAC protocol is not fixed and is affected by its surrounding environment and the data transmission and reception of neighboring nodes; its value should be less than 20ms. Simulations are then performed in a simple multi-hop chain topology and a randomly distributed topology. To study and verify the performance of the improved S-MAC protocol in a relatively reliable environment, we first tested it in a simple multi-hop chain topology consisting of 11 nodes. The distance between neighboring nodes was configured to 100 meters. The simulation results are shown in Figure 2: [align=center] Figure 2.1 Energy Analysis Figure 2.2 Delay Analysis[/align] Figure 2.1 illustrates the comparison of energy overhead between the S-MAC protocol and the improved S-MAC protocol in a multi-hop chain topology. The energy overhead of both MAC protocols increases with the number of hops. Figure 2.1 shows that when the number of hops increases to 5, the improved S-MAC protocol saves almost half the energy compared to the original S-MAC. This is because the activity time of the improved S-MAC protocol varies depending on the amount of traffic, while the original S-MAC protocol sends data packets to some nodes that are not the next hop, occupying extra activity time and wasting energy. To verify the improved S-MAC protocol's ability to reduce node sleep latency, we measured the end-to-end latency of data packets under light load conditions. Figure 2.2 shows the simulation results under different hop counts. The improved S-MAC protocol shows a significant improvement in latency compared to the original S-MAC protocol, especially after 4 hops, where the improved S-MAC protocol reduces latency by approximately 60%. Because the interleaved scheduling mechanism enables data to be continuously transmitted from the source node to the destination node, the transmission delay caused by node sleep is reduced. The improved effect is shown below. 50 nodes are randomly configured in a 100*500m² area. We change the number of source nodes by randomly selecting nodes from the network edge. All source nodes generate one message every 3 seconds. [align=center] Figure 3.1 Energy Analysis Figure 3.2 Delay Analysis[/align] Figures 3.1 and 3.2 analyze the S-MAC protocol and its improved protocol in terms of energy and delay, respectively. This is a simulation of the protocol under a more realistic and complex environment. As can be seen from Figure 3.1, the energy consumption of the original S-MAC protocol increases linearly with the increase in the number of source nodes, while the energy consumption curve of the improved S-MAC protocol is approximately a smooth straight line, around 500 joules. Figure 3.2 shows a comparison of the S-MAC protocol delay before and after the improvement. As the number of source nodes increases, interference increases, leading to a continuous increase in the original S-MAC delay, reaching 2.7 seconds with 36 source nodes. The improved S-MAC protocol, however, suffers less interference due to its ability to transmit data continuously, with a maximum delay of only 1.6 seconds. 6. Conclusion This paper provides an in-depth analysis of the S-MAC protocol in wireless sensor networks, pointing out its existing problems. Then, combining the T-MAC and D-MAC protocols for wireless sensor networks, it proposes its own improvement method. Simulation analysis shows that this method effectively reduces energy consumption caused by node idle listening and alleviates the transmission delay problem caused by node sleep. The improvement effect is significant and has practical implications. References [1] Akyildiz LF, Su WL, Sankarasubramaniam Y, Cayirci E. A survey on sensor networks[J]. IEEE Communications Magazine, 2002, 40(8):102～114. [2] Ye W, Heidemann J, Estrin D. An energy-efficient MAC protocol for wireless sensor network[C]. In: Proc 21st Int'l Annual Joint Conf IEEE Computer and Communication Societics (INFOCOM 2002), New York, NY, June 2002 [3] IEEE802.11 Wireless LAN Medium Access Control and Physical Layer Specifications[S], 1997 [4] Yuan Honglin, Xu Chen, Zhang Guoan. TOSSIM: Wireless Sensor Network Simulation Environment[J]. Microcomputer Information, 2006, 7-1: 154-156