Impact of Security Threats on IoT Telecommunication Network Performance

The integration of Internet of Things (IoT) devices within telecommunication networks has led to the emergence of many security vulnerabilities affecting network performance, reliability, and information integrity. In this regard, this paper provides a comprehensive analysis and evaluation of the effect of cyber threats and cybersecurity solutions for IoT-based telecommunication networks as well as providing an efficient solution. The impact assessment carried out indicates that attacks such as DDoS and MITM attacks significantly impair the network's efficiency since they result in reduced network throughput of 50 Mbps, increased latency to 300 ms, up to 30% packet loss, and energy consumption of up to 80 J. Current approaches used to enhance security include firewalls and encryption algorithms like AES, with the latter achieving a detection accuracy of 98%, but at the expense of high energy consumption of 10 J, while the former achieves 90% detection rate at minimum overhead, but with high latency. An innovative feedforward neural network-based intrusion detection system (FFNN-IDS) was developed with impressive results of a detection accuracy of 99%, low latency of 18 ms, and zero false-positive rates compared to traditional security mechanisms. Post-implementation of the proposed solution resulted in significant improvement in performance, whereby throughput increased from 60 to 85 Mbps, decreased latency from 220 to 150 ms, and reduced packet loss from 25% to 10%, thus enhancing network stability, efficiency, and performance. There was also a decrease in the incidence of cyber threats, such as unauthorized access reduced to 5% from 50% daily, DDoS attacks were reduced to 20% from 80%, whereas the frequency of network downtime was greatly reduced from 120 minutes per day to 20 minutes per day. The efficiency of the proposed solution is further evidenced by the significant increase in data integrity from 65% to 95% despite increased energy consumption of 90 J. Overall, it was observed that the proposed approach performed efficiently despite increased processing overhead as the scale increases from 5% to 80% as the number of devices increases from 100 to 50,000.