Somia Sahraoui

Massive volumes of data generated by large numbers of connected Internet of Things (IoT) devices impact both IoT service and network performance. Packet routing is an ever-important networking feature that enables the transmission of data packets across interconnected IoT networks. Usually, classical proactive and reactive routing protocols use the information of their neighbours to build a global view of the network by the mean of control messages and then calculate a path to the destination. After a packet has been delivered, history is discarded, and only actual information is used to determine new paths. In the present research work, we propose applying a Machine Learning (ML) approach to take advantage of previously successfully taken routing decisions and build new decision models that nodes can use to make intelligent routing decisions. In our proposal, we integrate three machine learning techniques: Decision Trees(DT), Support Vector Machines (SVM), and Artificial Neural Networks (ANN) into the well-known Ad hoc On-demand Distance Vector (AODV) routing protocol and evaluate the efficiency of solution variants for packets routing in IoT-connected smart city scenario. The assessment results show that the proposed ML-based routing solutions outperform the AODV protocol and decrease the use of control messages to more than 50% of the overall communication overhead. Therefore, network performances are more optimized, and network lifetime is also increased.

Diabetic patients are prone to daily and severe health‐related risks, namely hyper and hypoglycaemia. Hypoglycaemia phenomenon happens when the glucose level in patient's blood is lower than a well‐determined sill. It may induce serious impacts, such as functional brain failure or even the death. Hypoglycaemia is especially dangerous when it occurs during the night while the patient is asleep because it becomes difficult to be detected by the patient itself or other persons around him. While all existing sensor‐based solutions are detection‐only driven, the proposed solution goes beyond and attempts to treat autonomously, and at low cost, the nocturnal hypoglycaemia. The presented system detects the nocturnal hypoglycaemia phenomenon based on accelerated heart‐rate symptom and a progressive detection algorithm. The system treats then the detected nocturnal hypoglycaemia throughout safe and automatic injection of glucagon.

We deal with privacy problems resulting from the possibility of communication-flow identification and tracking in the currently designed IPv6 Low Power Wireless Personal Area Network (6LoWPAN)-based solutions for the Internet of Things (IoT). We propose a solution that makes it difficult for an attacker to detect that a communication is taking place between a source and a destination entities by disguising the main flow identifying fields in 6LoWPAN packet headers. Our solution is based on the One Time Address (OTA) approach proposed for the traditional Internet, with a focus on low complexity, memory footprint, and energy consumption. We evaluated our solution that we called micro OTA (uOTA) by implementing it on the Cooja simulator and compared it with micro Moving Target IPv6 Defense (uMT6D). Results show that our solution improves communication flow privacy in IoT and outperforms uMT6D in energy consumption, network overhead, latency, and privacy preservation.

Ensuring security for communications, services, devices, and end‐users has been and continues to be a major concern in today's Internet. Recently, with the growing research advances toward the realization of the Internet of Things (IoT), security concerns are expected to rise substantially. This is due to the fact that a sheer number of tiny wireless and smart devices, handling personal and contextual data, will be connected permanently to the Internet. The connected things are henceforth able to interact autonomously with each other and with their users. This paper focuses on Human‐to‐Thing (H2T) interactions that are necessary in many important IoT applications, namely E‐healthcare, connected home, smart cities, and so on. Such a communication style is characterized by many technological and material heterogeneities, which exposes the connected (and already constrained) things to Denial of Service (DoS) threats and makes security more challenging. This paper addresses the security issue with H2T communications and presents an asymmetric‐selective mechanism to guarantee effective protection of such communication style while ensuring good resilience against DoS attacks. The assessment results show the efficiency of our proposal.

Wireless Sensor Networks (WSNs) are widely and successfully employed in various application domains. They are easily deployed to collect valuable information and monitor potential environmental phenomena. However, the special nature of WSNs as well as their severe constraints and resource limitations make them vulnerable to various types of threats. Replay attack, is one example. According to this attack, the adversary intercepts and replays several times the same (old) message leading either to missed alerts or to false alerts. Many solutions have been proposed to mitigate message replay attack. However, all these solutions are of cryptographic natures and consider only external attacks exercising a trivial scenario of replay attack. In fact, the attacker could be a lot smarter, and in this case, it replays only the data field in the message while keeping the remaining fields updated. This novel form of replay attack is much more dangerous and difficult to be detected. We call this attack variant by data replay attack. As sensor nodes may be easily captured and compromised, the worst scenario occurs if data replay attack is performed by an internal intruder. In this paper we propose an efficient intrusion detection framework to overcome data replay attack in WSNs. The proposed intrusion detection system is named DR-IDS (Data Replay Intrusion Detection System). The performance evaluations performed under NS2 simulator show that the proposed solution is sufficiently robust.