The sharing of sensitive medical data among institutions presents a significant challenge due to strict privacy regulations, the need for robust de-identification processes, and the ethical imperative to protect patient confidentiality. Federated Learning (FL) addresses these challenges by enabling institutions to collaboratively train AI models on decentralized data, thereby enhancing privacy and security without directly sharing sensitive patient information. However, FL requires complex synchronization implementations, has costly communication overheads, and may fail to capture data heterogeneity across institutions. In this work, we propose Personalized Mixture of Local Experts (P-MoLE), a Personalized Federated Learning (PFL) approach that effectively combines predictions from multiple locally trained models in a sample-specific manner. Leveraging both the individuality of each local model and variation across the ensemble, P-MoLE learns the profile of each institution's local model and strategically weighs their prediction's contributions to the final segmentation. This approach harnesses the heterogeneity of each institution 's unique data to increase the generalization capabilities across all institutions. By each institution sharing only the final models trained locally on the sensitive data, no private patient data is exposed and the need for expensive communication infrastructure is removed. Results across two popular multi-institutional medical imaging datasets show P-MoLE achieves state-of-the-art performance without the extensive cooperative effort requirement of previous works. Additionally, ablation study results show that P-MoLE is flexible to the number of local models in the ensemble, increasing performance over the local models alone in each case.

Modern cellular networks are complex systems offering a wide range of services and present challenges in detecting anomalous events when they do occur. The networks are engineered for high reliability and, hence, the data from these networks is predominantly normal with a small proportion being anomalous. From an operations perspective, it is important to detect these anomalies in a timely manner in order to mitigate them and preclude the occurrence of major failure events. In telecommunications diverse set of data such as KPIs, logs and alarms are generated to monitor the health and stability of the network element and the services carried over it [1]–[4].

In this paper, we address the detection of face-to-face social interactions between people using unobtrusive, wearable iBeacon devices, as well as the determination of groups and communities that these individuals form. This is a challenging problem since this type of sensor data is very noisy, often incomplete with many missing values, and easily perturbed by motion and nearby obstacles, especially in very dynamic indoor environments. The key idea of our approach is to transform the original noisy sensor data into specific feature vectors by a series of statistical methods designed to overcome the challenges in the data. These features vectors, reflecting the interaction between people, are then clustered to reveal communities without the need to rely on localization information or pre-defined interaction models and parameters. Our data-driven approach provides robust community detection that is not sensitive to noise and missing signals, and automatically captures the dynamic interaction between people. Our approach is scalable for application in real-time.

Mobile phones have become one of the primary tools for individuals to communicate, to access data networks, and to share information. Service providers collect data about the calls placed on their network, and these calls exhibit a large degree of variability. Providers model the structure of the relationships between network subscribers as a mobile call graph. In this paper, we apply a new measure to quantify by how much a relationship between users in a mobile call graph deviate from an average relationship. This measure is used to explore the connection between calling behaviors and the complex structure mobile call graphs take. We study a large call graph from a major service provider and learn that distant, outlier relationships play the largest role in maintaining connectivity between cellular users, and that calling features of users more strongly influence tie variation compared to social features. We also observe a rapid decay of its massively connected component as outlier ties are removed.

MPEG-based compression schemes and multicast- based nature of IPTV deployments result in channel change delay. Solutions to reduce this delay have been proposed in the literature and also in the commercial deployments. In this paper, we present an analysis of channel change delay in multicast-based IPTV systems. As video encoding parameters, more specifically the frequency of the intra-coded (I-) frames, have major impact over the channel change delay, we also investigate how different intra-coded frame frequency affects the channel change delay. In the second part of the paper, we evaluate the bandwidth penalty of using more frequent I- frames to reduce the channel change delay. Our results show that, maximum channel change delay can be reduced to around 830 msec with acceptable bandwidth penalties without using an explicit channel change acceleration mechanisms. When the delay requirement is stricter or the delay variation is not acceptable, an explicit channel change acceleration mechanism will be necessary.

Critical national infrastructures for transportation, power, finance, and other basic industries rely heavily on information and telecommunications networks (voice, data, Internet) to provide services and conduct business. While these networks tend to be highly reliable, disasters may lead to extended outages requiring days/weeks to repair. These outages cause loss of business continuity and financial transaction failures. This paper describes a dynamic simulation model of communications network disasters, their network performance, and their impact on other critical infrastructures.

The recent reduction in telecommunications spending has increased the importance of network planning to improve the return on investment on the existing network infrastructures. Therefore, tools that help in maximizing the bandwidth efficiency of the network at a minimum cost are essential. Previous work in this area focused on increasing bandwidth efficiency and reliability. In this work, in addition to increasing the bandwidth efficiency, we address the complexity of network management and operations. This issue is explicitly addressed by our novel framework, a simple polynomial time algorithm (SimPol) that achieves optimum network performance (in terms of congestion or bandwidth consumption) using only a small number of paths. The problem formulation is based on splittable multicommodity flows. Using SimPol we show that the total number of paths is at most k+m, where k and m are the numbers of demands and edges in the network, respectively. We extend the basic framework into an integer programming formulation to address the tradeoff between network congestion and the total number of paths. We also use SimPol to address the problem of implementing path/link policies such as bandwidth-limited paths. The performance of SimPol is evaluated through extensive simulations. Using the integer programming approach, we can get exactly one path while losing about 10% to 50% in congestion depending on the number of demands. This congestion is, however, far better than the traditional shortest path routing. The framework is general and can be used in capacity planning for transport networks such as MPLS and ATM.

The geographical connection admission control (GCAC) algorithm is introduced for low Earth orbit (LEO) satellite networks. The GCAC scheme estimates the future handover blocking probability of a new call attempt based on the user location database, in order to decrease the handover blocking. By simulation, it is shown that the proposed GCAC scheme guarantees the handover blocking probability to a predefined target level. Since GCAC algorithm utilizes the user location information, performance evaluation shows that this technique also guarantees the target level of handover blocking probability in the nonuniform traffic pattern.

In future IP networks, nodes supporting new congestion avoidance mechanisms like RED and ECN are going to exist together with nodes carrying traditional drop tail buffers. We study the performance of a set of TCP flows passing through two congested nodes in such a heterogeneous network. Our results show that unequal hop behaviour (in the form of RED and drop tail) in the congestion points can increase the bias against TCP flows passing through multiple congested nodes. Parameter settings of RED become critical again in the dual-bottleneck situation just like in the single-bottleneck case. Furthermore, appropriate settings for single- and dual-bottleneck situations appear to be significantly different, making it more problematic for RED to provide a "better-than-drop tail" performance with a fixed set of parameters.

In this paper, a connection admission control algorithm, referred to as the geographical connection admission control (GCAC) algorithm, is introduced to handle the spot-beam handovers in the low Earth orbit (LEO) satellite networks. Spotbeam handovers occur frequently because of the small spotbeams and high satellite speed. The GCAC algorithm admits a new call into the network only if the expected handover call blocking probability of the new and the existing calls is not increased beyond a system threshold. The algorithm utilizes the deterministic satellite movement pattern and the user locations to estimate the expected handover call blocking probability. The experimental results show that the algorithm limits the handover call blocking probability and can handle nonuniform traffic distribution over the coverage area of the network.

This paper describes current and proposed protocols for mobility management for public land mobile network (PLMN)-based networks, mobile Internet protocol (IP) wireless asynchronous transfer mode (ATM) and satellite networks. The integration of these networks will be discussed in the context of the next evolutionary step of wireless communication networks. First, a review is provided of location management algorithms for personal communication systems (PCS) implemented over a PLMN network. The latest protocol changes for location registration and handoff are investigated for mobile IP followed by a discussion of proposed protocols for wireless ATM and satellite networks. Finally, an outline of open problems to be addressed by the next generation of wireless network service is discussed.

Low Earth orbit (LEO) satellite networks have dynamic, yet deterministic, topologies. The time-varying connectivity pattern would result in the re-routing of all connections passing through a link that is turned off as a result of the topology change. A routing algorithm called probabilistic routing protocol (PRP) is introduced. The PRP reduces the number of re-routing attempts due to the dynamic topology of the network. During the routing phase of a newly arriving call, the PRP eliminates the links that will be turned off before the call releases the link due to call termination or connection handover. Since the algorithm has no knowledge of the call duration or exact terminal location, route usage time is only known probabilistically. The probability distribution function of the route usage time of the call is determined to realize the algorithm. Since the routing algorithm works in parallel with a handover re-routing algorithm, the application to the footprint handover re-routing protocol (FHRP) is also demonstrated. The performance of the algorithm is investigated using simulation experiments.

Low Earth Orbit (LEO) satellite networks require a reliable handover re-routing protocol that is critical for connections with multihop intersatellite links (ISLs). In this paper, we introduce a footprint handover re-route protocol (FHRP) that maintains the optimality of the initial route without performing a routing algorithm after satellite handovers. Furthermore, the FHRP handles the inter-orbit handover problem. The FHRP makes use of the footprints of the satellites in the initial route as the reference for re-routing. More specifically, after an optimum route has been determined during the call establishment process, the FHRP ensures that the new route due to handover is also optimum. The FHRP is applicable to any type of connectionoriented networks. The adaptation of the FHRP to the widely accepted ATM technology, which will be employed in future satellite networks, is also addressed in this paper. The FHRP demands easy processing, signaling, and storage costs. The performance re...