Hybrid IoT communication approach

As depicted in Fig. 2, the system integrates embedded modules for sensing, processing, and communication, while also deploying logical networking overlays and protocol stacks to support stable operation across wired and wireless channels. Key software integrations include the TI 15.4 stack for RF communication and a VPN-based solution for creating a logical mesh network over the PLC infrastructure. Additionally, this section outlines the technical decisions behind hardware–software compatibility, the configuration of network roles, and the performance validation setup. The overall system is designed to maintain high data availability and recover from link failures by intelligently switching between communication mediums based on real-time network conditions.

Fig. 2: Hardware architecture.

RF network

The RF communication channel was thoroughly tested to validate its performance, connectivity, and seamless integration with the central DCU and backend infrastructure. The setup employed TI CC1352R modules operating under the TI 15.4 stack, configured in a star topology, as previously described. To assess the signal quality and transmission range, a controlled point-to-point test was conducted between two RF nodes in a typical indoor lab environment at a varying distance of 0.5 meters to 1.0 meters. In this experiment, one node continuously transmitted data packets while the receiving node measured the Received Signal Strength Indicator (RSSI) at varying power levels. The evaluation, illustrated in Fig. 3, shows the relationship between transmission power and the resulting RSSI. While higher power levels did yield marginal RSSI increases, the performance gains plateaued significantly after 10 dBm. Therefore, the 10 dBm level was selected as the optimal operating point. This power level provides a strong RSSI (approx. -40 dBm), ensuring a reliable link margin, while avoiding the diminishing returns and unnecessary energy consumption of operating at maximum power.

Fig. 3: RSSI level corresponding to tested power.

Following the initial range validation, three sensor nodes were deployed, each built around an STM32F401RCT6 microcontroller connected to a CC1352R RF module. These nodes were configured as child devices under the TI 15.4 protocol stack and successfully established connections with a single RF DCU node. The sensor nodes were programmed to periodically transmit structured packets containing unique sensor identifiers and simulated (dummy) data values. The DCU’s monitor showed the correct number of joined devices (3), which confirmed successful reception of the transmitted packets, and then it forwarded the aggregated data to the cloud-based web server for further processing and visualization. This test validated end-to-end communication, demonstrating the correct functioning of both RF data acquisition and backend data integration.

PLC network

For the PLC network, a mesh topology was logically implemented to allow flexible multi-hop data forwarding among smart meters. However, it is important to note that the physical topology as illustrated in Fig. 4 remains a bus configuration, as all nodes share a single electrical backbone originating from the utility provider’s power line, and then different nodes connect to it. This structure enables cost-effective deployment by leveraging existing infrastructure, while the mesh logic ensures improved connectivity and adaptability in data routing, particularly in scenarios with disrupted direct communication paths. Although the physical topology of the PLC network follows a bus configuration, with all nodes connected to a single power line backbone from the utility provider, it operates as a logical mesh. In this design, each node is aware of its peers and can establish direct or indirect communication paths as needed. This is especially beneficial for long-range deployments spanning several kilometers, where some nodes may face signal degradation or be unable to reach the destination directly. Multi-hop forwarding addresses this by allowing packets to be relayed through intermediate nodes until they reach the DCU. As a result, the logical mesh significantly improves communication success rates and link redundancy in practical PLC environments.

Fig. 4: PLC physical topology.

The PLC network was validated in two phases using commercially available LX200V50 modules as a proof of concept to test its logical capabilities. First, a point-to-point (P2P) test was conducted to establish a stable communication range, which was determined to be approximately 10 meters within the laboratory environment. Second, a 5-node logical mesh test was performed to validate the system’s functional operations, such as the VPN overlay and fault-tolerant rerouting. In this test, five PCs simulated smart meter nodes on the same electrical phase, with nodes spaced approximately 1.2 meters apart and a 2-meter separation between Node3 and Node4. This configuration was designed to verify the mesh’s logical self-healing capabilities, which are distinct from the physical signal performance over a large-scale distribution network. A VPN overlay created a logical full mesh, allowing secure, direct communication between all nodes. To test fault tolerance, the direct logical link between Node1 and Node3 was administratively disabled, demonstrating the system’s ability to autonomously reroute traffic through alternative paths (via Node2, for example).

Fig. 5: Validating data forwarding from PC2.

However, with the link disabled, the system relied on the VPN’s dynamic routing capabilities, which successfully rerouted traffic through PC2. This intermediate relay behavior was confirmed using network monitoring tools, as shown in Fig. 5, where relayed packets were observed on PC2. The successful forwarding of data in the absence of a direct path demonstrated the self-healing nature of the logical PLC mesh network, validating its suitability for smart metering applications in variable and failure-prone electrical environments.

Following the successful topological validation of the PLC mesh, the network’s link quality was assessed using key performance metrics such as latency, packet loss, and throughput. These measurements were collected across different PC pairs within the mesh under typical operating conditions. The evaluation aimed to quantify the communication reliability and efficiency of the LX200V50-based PLC links under mesh routing. The results, summarized in Table 2, provide insight into the variability of link performance and the effectiveness of the mesh topology in maintaining stable data transmission across diverse electrical paths.

Test | Bitrate (Mbits/sec) | Jitter (ms) | Lost/Total datagrams | Loss (%)
Test 1 | 21.4 | 1.812 | 17,955 / 134,203 | 13
Test 2 | 22.3 | 0.445 | 14,362 / 134,195 | 11
Test 3 | 23.9 | 0.871 | 5,558 / 134,214 | 4.1

While this initial point-to-point and mesh evaluation recorded packet losses between 4% and 13%, it is important to note that these figures stem from prototype implementation overheads in early firmware rather than fundamental media limits, and would typically be mitigated in a production AMI setup (which requires >99% reliability) through ARQ mechanisms, robust error-correction coding, and optimized hardware drivers. Nonetheless, this test validates the logical feasibility and self-healing nature of deploying PLC as a communication backbone in a distributed mesh network.