The increasing availability of AI-based tools such as ChatGPT has transformed how students conduct online searches. This study investigates how students search for information with and without generative AI assistance, how distinct behavioural patterns emerge, and how these patterns relate to performance outcomes. We analyse 726 search stories, i.e., their navigation actions, from 305 students completing up to three search tasks. We first examine performance differences across search modalities, comparing AI-assisted and traditional web search approaches. We then analyse students’ search behaviours and examining their association with performance outcomes. Preliminary results show that AI-based searching tends to cluster around specific behavioural combinations and reduces extreme errors, but does not systematically improve correctness compared to traditional web searches. Furthermore, while performance outcomes vary across different search behaviours, no consistent patterns emerge linking specific behaviours to stable performance across tasks.

Proprietary public APIs play a crucial and growing role as research tools among social scientists. Among such APIs, Google's machine learning-based Perspective API is extensively utilized for assessing the toxicity of social media messages, providing both an important resource for researchers and automatic content moderation. However, this paper exposes an important bias in Perspective API concerning German language text. Through an in-depth examination of several datasets, we uncover intrinsic language biases within the multilingual model of Perspective API. We find that the toxicity assessment of German content produces significantly higher toxicity levels than other languages. This finding is robust across various translations, topics, and data sources, and has significant consequences for both research and moderation strategies that rely on Perspective API. For instance, we show that, on average, four times more tweets and users would be moderated when using the German language compared to their English translation. Our findings point to broader risks associated with the widespread use of proprietary APIs within the computational social sciences.

Model calibration and feature selection are two critical aspects in developing reliable and accurate machine learning models. Calibration ensures that the model's confidence scores accurately reflect the true likelihood of correctness, which is essential in security-critical applications, like Network Intrusion Detection Systems (NIDS). Feature selection (FS), meanwhile, enhances model efficiency, interpretability, and generalization by identifying the most relevant inputs. However, it may also degrade model's calibration if not explicitly considered. Recently, Explainable Artificial Intelligence (XAI) methods, particularly Shapley Additive Explanations (SHAP), have proven effective in guiding the FS process. Yet, existing SHAP-based FS techniques typically focus on improving accuracy, often giving little attention to the impact of FS on model calibration, an aspect that is critical for reliable decision-making in high-stakes applications such as NIDS. In this work, we propose a novel approach that incorporates a calibration-aware loss function for XGBoost with SHAP-based Recursive Feature Elimination for FS to jointly improve both predictive accuracy and model calibration. Experiments on two benchmark NIDS datasets show that our approach can reduce Brier score and Expected Calibration Error by up to 3.5% and 84.8%, respectively, over uncalibrated baselines, and by up to 2.6% and 58.3% over standard calibration methods, while enhancing or maintaining predictive accuracy and number of features.

The aftermath of the 2020 US Presidential Election witnessed an unprecedented attack on the democratic values of the country through the violent insurrection at Capitol Hill on January 6th, 2021. The attack was fueled by the proliferation of conspiracy theories and misleading claims about the integrity of the election pushed by political elites and fringe communities on social media. In this study, we explore the evolution of fringe content and conspiracy theories on Twitter in the seven months leading up to the Capitol attack. We examine the suspicious coordinated activity carried out by users sharing fringe content, finding evidence of common adversarial manipulation techniques ranging from targeted amplification to manufactured consensus. Further, we map out the temporal evolution of, and the relationship between, fringe and conspiracy theories, which eventually coalesced into the rhetoric of a stolen election, with the hashtag #stopthesteal, alongside QAnon-related narratives. Our findings further highlight how social media platforms offer fertile ground for the widespread proliferation of conspiracies during and in the aftermath of major societal events, which can potentially lead to offline coordinated actions and organized violence.

Emotions play a significant role in product design for end-users.However, how to take emotions into account is not yet completely understood.We argue that this gap is due to a lack of methodological and technological frameworks for effective investigation of the elicitation conditions related to emotions and corresponding emotional responses of the users.Emotiondriven design should encompass a thorough assessment of users' emotional reactions in relation to certain elicitation conditions.By using Virtual Reality (VR) as mean to perform this investigation, we propose a novel methodological framework, referred to as the VR-Based Emotion-Elicitation-and-Recognition loop (VEE-loop), to close this gap.

The problem of optimally deploying a virtual network onto a substrate physical network is referred to as Virtual Network Embedding (VNE). In general, this embedding is requested by a customer to an Internet Service Provider (ISP), which performs the VNE over its physical telecom network. In several situations, the physical substrate infrastructure is composed of multiple independent ISPs. In this scenario, ISPs are concerned about exposing to a third-party entity (e.g., the customer) sensitive infrastructural details that are needed to perform an effective embedding. Following a common privacy-preserving approach, known as Limited Information Disclosure (LID), the embedding may be performed by the customer based on a limited and abstracted view of the multi-domain infrastructure that ISPs accept to expose. With this approach, embedding is sub-optimal (e.g., embedding cost is not minimized) in comparison with the case where all information is available, i.e., Full Information Disclosure (FID). In this work, we propose a Reinforcement-Learning-based algorithm able to process data that the customer and ISPs cipher under the Shamir Secret Sharing (SSS) scheme. This approach guarantees total privacy to both the customer and the ISPs (e.g., details about a virtual function are only revealed to the ISP in charge of hosting it) and achieves comparable embedding cost of an existing FID heuristic, as observed from extensive simulations. The main drawback of our algorithm is the high overhead of data that ISPs and the customer need to exchange with each other to execute it. Hence, we also explore the trade-off between embedding cost and data overhead resulting from the reduction of operations done by the RL. In general, intermediate embedding costs between the FID and LID heuristics can be obtained at a significant reduction of data overhead, while not sacrificing any privacy guarantees.

Content Providers (CPs) typically encrypt the content sent over the telecom network to improve security and privacy of their final users, as well as to protect business-critical information (e.g., contents' popularity). Due to this encryption, Internet Service Providers (ISPs) can not easily apply caching strategies that require the inspection of traffic traversing their networks to select the most popular contents. The most common approach to solve the conflict between privacy and caching consists in allowing a CP to manage the caches (e.g., by storing and delivering the contents) directly from inside the area of the ISP. However, in this way ISPs lose the legitimate control on a portion of traffic traversing their networks. An alternative approach is enabled by recently-proposed architectural solutions that allow a CP to encrypt the contents and associate pseudonyms to them, and the ISP to count the occurrences of such identifiers to infer popularity-related information without inspecting the original contents. However, we observe that ISPs can still obtain valuable information about contents' popularity that may threaten CPs' privacy. In this paper, we formalize a strategy of association between pseudonyms and contents that effectively improves privacy but leads to a degradation of caching performance. We formally define privacy in this context and study the trade-off between caching and privacy considering differerent metrics, such as the hit-rate and the retrieval latency. The results, obtained by means of simulations over both real and synthetic data, show that privacy can be significantly improved while accepting a minor impact on the hit-rate of caching and suggest the applicability of the considered architecture in a real scenario of content delivery.

The distribution of video contents generated by Content Providers (CPs) significantly contributes to increase the congestion within the networks of Internet Service Providers (ISPs). To alleviate this problem, CPs can serve a portion of their catalogues to the end users directly from servers (i.e., the caches) located inside the ISP network. Users served from caches perceive an increased QoS (e.g., average retrieval latency is reduced) and, for this reason, caching can be considered a form of traffic prioritization. Hence, since the storage of caches is limited, its subdivision among several CPs may lead to discrimination. A static subdivision that assignes to each CP the same portion of storage is a neutral but ineffective appraoch, because it does not consider the different popularities of the CPs' contents. A more effective strategy consists in dividing the cache among the CPs proportionally to the popularity of their contents. However, CPs consider this information sensitive and are reluctant to disclose it. In this work, we propose a protocol based on Shamir Secret Sharing (SSS) scheme that allows the ISP to calculate the portion of cache storage that a CP is entitled to receive while guaranteeing network neutrality and resource efficiency, but without violating its privacy. The protocol is executed by the ISP, the CPs and a Regulator Authority (RA) that guarantees the actual enforcement of a fair subdivision of the cache storage and the preservation of privacy. We perform extensive simulations and prove that our approach leads to higher hit-rates (i.e., percentage of requests served by the cache) with respect to the static one. The advantages are particularly significant when the cache storage is limited.

Content delivery involves multiple entities, such as Content Providers (CPs) and Internet Service Providers (ISPs). To better serve its users, the CP may deploy resources (e.g.,caches) as close as possible to them. In this work, we consider the deployment of Virtual Servers (VSs) to stream live videos owned by the CP in the network of the ISP. An efficient deployment requires the knowledge of both the users' position and requests. However, the CP knows what users request but not their exact position, while the ISP has knowledge of users' locations but not of their requests (due to content encryption). To guarantee users' privacy, the ISP and the CP cannot exchange these information with each other. In this paper, we make the two parties cooperate by employing a secure multiparty computation protocol which does not require the two parties to reveal the aforementioned information. This protocol allows the ISP to obtain the number of requests for a specific live-video content issued from a given area at the cost of a negligible overhead. Knowing this information, the ISP efficiently deploys the VSs with the aim of minimizing the number of hops crossed by the live videos to reach their viewers. We assess the average number of hops saved when the geographic distribution of requests is known and we conclude that it is relevant. Then, we investigate scenarios in which the privacy of both the CP and the ISP can be violated, and we propose several countermeasures. In particular, the parties can distort their data when executing the protocol, which results in a trade-off between performance and privacy. We conclude that the fulfillment of stringent privacy requirements comes at significant performance loss.

In this paper, we present INDIGO, an interest-driven data dissemination framework that enables to predict the performance of data dissemination. INDIGO computes a tight upper bound of data dissemination time under (i) heterogeneous mobility patterns of people and, (ii) interests-driven dissemination strategy. INDIGO automatically learns the interests of people from their browsing history and also captures their heterogeneous mobility patterns.

Wearable devices have recently gained a foothold in the market with the uptake of smartwatches. The strong tie between a smartwatch and its owner, the highly predictable position of a smartwatch on the body, and its internal sensors are enabling a wide array of applications that leverage the user context. In this paper we focus on a gesture recognition system to augment the user interaction with multimedia applications. We define a set of seven gestures that are relevant across several applications and we collect an extensive dataset with two smartwatches (the Motorola Moto360 and Apple's Watch). We use Long Short Term Memory neural networks for gesture recognition based on sensor data from both smartwatches. We provide an extensive evaluation of the classification accuracy of the system and provide a sensitivity analysis to find the Long Short Term Memory configuration that maximizes the classification accuracy. We also show the extent to which Long Short Term Memory neural networks outperform traditional machine learning approaches. We also illustrate an application we built for the Android and iOS platforms that allows developers to easily integrate the gesture recognition in their own systems. We conclude the paper with a description of use cases to underscore the potential impact of our contribution.

Computational offloading is a traditional approach to augment the capabilities of smart devices. It has been shown that offloading is most efficient when carried out within a Local Area Network. Several frameworks have been proposed to tackle this problem but fail to consider that some of the application code may be part of the offloading request. This is a key element in cloudlets where it is virtually impossible to load all the necessary application code a priori on all nodes. It is even more crucial in opportunistic computing where it cannot be known a priori which devices will become the offloading targets.

To explore the applicability of mobile smart devices to personal protection against violent crime, we propose a system that can detect the onset of hazardous situations involving violent crime by leveraging standard activity recognition strategies on smartphones and sensory inputs from wearable devices, as well as send help requests to alert the authorities.

This paper presents a novel authentication strategy for Bluetooth-equipped smartwatches. We use the built-in smartwatch sensors to detect whether two users have shaken hands. If this is the case each devices give to the other a soft authentication privilege, which is suitable for applications with relaxed security needs (for instance, an application to exchange business cards). We evaluate the system using different machine learning techniques and we investigate their performance in the dimensions of accuracy and energy consumption.

Asynchronous low-power listening techniques reduce the energy footprint of radio communication by enforcing link layer duty cycling. At the same time, these techniques make broadcast traffic significantly more expensive than unicast traffic. Because broadcast is a key network primitive and is widely used in various protocols, recently several techniques have been proposed to reduce the amount of broadcast activity by merging broadcasts from different protocols. In this paper we focus on collection protocols and investigate the more extreme approach of eliminating broadcast completely. To this end, we design, implement and, evaluate a Broadcast-Free Collection Protocol, BFC. We derive first-order models to quantify the costs of broadcasts, and evaluate the performance of BFC on a public testbed. Compared to the Collection Tree Protocol, the de facto standard for data collection, BFC achieves double-digit percentage improvements on the duty cycles. The specific benefits to individual nodes depend on the relative cost of unicast activity; we show that the nodes that benefit the most are the sink's neighbors, which are crucial for network lifetime extension. Eliminating broadcast also brings several other advantages, including extra flexibility with link layer calibrations and energy savings in the presence of poor connectivity.

Wireless sensor networks (WSNs) penetrated the market mainly as solutions for specific application scenarios. However, this strong specialization limits WSNs reuse both in terms of development as well as in terms of technical results: Every new application scenario requires a new design, development and validation, as well as management skills. This is frustrating for any WSNs user, developer or manager. To reverse this tendency and thus improve the quality of experience and user-friendliness in WSNs, we designed FLEXOR, a sustainable software architecture optimized to support the implementation, rapid prototyping, evaluation, and testing of wireless sensor network applications, that is platform independent and user-friendly. FLEXOR is designed to accommodate many different applications and services for wireless sensor networks and foster code re-usability and cross-platform component re-usability. FLEXOR offers high modularity, well defined interfaces, remote node management functionality as well as run-time module exchange. Finally, the introduction of a unifying way for WSNs development opens to a higher homogeneity and thus to more easy comparison among different solutions. We present here an analysis of FLEXOR from these new angles and show how effective it is for several purposes and in particular for non-experts and in education.

The increasing spreading of top-medium range mobile phones offering advanced capabilities and the broader use of short-range wireless technologies embedded in them are laying the foundations for the development of new added-value services. In this paper we present an architectural solution for delivering location-based services over heterogeneous mobile phones tailored to the preferences of individual users. We then discuss the performance of a solution for proximity marketing derived from the platform, in terms of localization accuracy and client-application usability.

In many-to-one sensor network routing, neighbor tables are commonly viewed as the centerpiece of the control plane and are typically used to allow nodes to recover from route breakage. Our results, however, indicate that table-free routing is a competitive strategy.

We propose DisCord, an opportunistic protocol for neighbor discovery and data exchange coordination in sensor networks. DisCord allows multiple low-end nodes within radio range to discover each other, self-organize into a network, and exchange data in an autonomous fashion. DisCord can successfully cope with node mobility and unstable network dynamics.

The performance of sensor network protocols is greatly affected by the network topology: the network layout, the link dynamics, and the sink placement. We propose the Expected Network Delivery (END), a protocol-independent collection metric that captures the impact of the topology on protocol performance. Our preliminary results with two collection protocols on various topologies on the MoteLab testbed show that the END metric enables a more systematic and topology-aware evaluation of protocol performance.

We present a prototype for a straightforward proximity marketing solution (PROMO) over heterogeneous smart phones tailored to the needs of individual users. The system is based on the collection of Wi-Fi fingerprints in an environment, and sends data to the users' smart phones based on their interests and their location. We believe that our system has high potential because, differently from existing similar solutions, it is highly customizable and works also in indoor environments. The market validity is further confirmed by the decision to fund the development of this prototype taken by the Swiss Innovation Promotion Agency (CTI-No 11556.1).

The 11 papers in this special issue on mission critical networking are divided into three categories: quality of service issues (three papers); security issues (four papers); and configuration and data collection issues (four papers).

Mobile communication is increasingly becoming part of everyday life. Haggle (www.haggleproject.org) is an innovative project that exploits human mobility for creating an invisible social network to enable communication in the presence of intermittent network connectivity. Mobile devices equipped with network interfaces, may in fact (temporarily) be out of range of the Internet infrastructure, and thereby unable to access Internet services. However, these devices may often be in range of other networked devices, or pass by an area where some type of connectivity is available (e.g. WiFi hotspots) and may be able to perform useful networking transactions involving local exchanges of data, exploiting, in this way, the opportunistic communication. One of the mechanisms for realizing the opportunistic communication is by taking advantage of context information (such as location) for correlating on-line social and physical networks. This, together with the need for information regarding devices locations and the dasiapervasivepsila presence of wireless networks in indoor environments, suggested us to implement a lightweight application able to localize mobile devices, exploiting opportunistic networking.

For low-power wireless links in rich scattering environments, multipath fading can easily dominate over the large-scale path loss. With the help of a link estimator that merges channel state information with data-driven feedback, it is possible to opportunistically leverage on positive fading states induced by human mobility to streamline routing in static low-end sensor networks.

To streamline the development of protocols and applications for wireless sensor networks with mobile elements, we propose the use of a virtual mobility overlay (ViMobiO). Using an idea from mobile ad hoc network research, physical nodes are seen as possible locations of logical nodes, whose motion is simulated with a state transfer between physical nodes. To maximize flexibility, state transfers are performed over the radio rather than over the wired backchannel, enabling testbed users to choose their mobility strategy within the constraints of network connectivity and to build on top of ViMobiO with no need to modify the testbed management software. Though the vagaries of low-power wireless propagation must be accounted for to ensure successful state transfers, our approach returns meaningful mobility profiles and adds a new dimension to large-scale experiments with wireless sensor networks.

We propose a new routing scheme (SpatioTempo) based on modeling the mobility of device carriers in periods and cycles. SpatioTempo exploits both the spatial (context) and the temporal (periodical behavior) characteristics of people to better route and optimize the message retransmissions. The combination with Propicman[1] imparts the strong points of the latter, such as context based routing and routing with zero knowledge.

Routing in intermittently connected mobile ad hoc networks (ICMAN) is a very challenging problem because disconnections are prevalent and the lack of knowledge about network dynamics hinders good decision making. In ICMAN one of the most important decisions is how to choose the most suitable intermediate(s) to forward the message to the destination. We propose in this document a new algorithm (PROPICMAN) that, based on context information, allows the sender to select the neighbor(s) such that the message has the highest probability to reach the destination.

Using location information to help routing is often proposed as a means to achieve scalability in large mobile ad hoc networks. However, location-based routing is difficult when there are holes in the network topology and nodes are mobile or frequently disconnected to save battery. Terminode routing, presented here, addresses these issues. It uses a combination of location-based routing (terminode remote routing, TRR), used when the destination is far, and link state-routing (terminode local routing, TLR), used when the destination is close. TRR uses anchored paths, a list of geographic points (not nodes) used as loose source routing information. Anchored paths are discovered and managed by sources, using one of two low overhead protocols: friend assisted path discovery and geographical map-based path discovery. Our simulation results show that terminode routing performs well in networks of various sizes. In smaller networks; the performance is comparable to MANET routing protocols. In larger networks that are not uniformly populated with nodes, terminode routing outperforms, existing location-based or MANET routing protocols.

This paper considers the problem of routing in a wide area mobile ad-hoc network called terminode network. Our routing scheme is a combination of two protocols called terminode local routing (TLR) and terminode remote routing (TRR). TRR is activated when the destination is remote and uses location of the destination obtained either via location management or by location tracking. TLR acts when the packet gets close to the destination. The use of TRR results in a scalable solution that reduces dependence on the intermediate systems, while TLR allows us to reduce problems due to location inaccuracy. The paper describes TLR and TRR protocols and presents some simulation results.

In this paper we simulated terminode routing. This routing scheme is designed for wide area networks, where a large part or all the nodes are mobile. Terminode routing is a combination of two protocols called Terminode Local Routing (TLR) and Terminode Remote Routing (TRR). TLR is used to route packets to close destinations. TRR is used to route to remote destinations and is composed of the following elements: Anchored Geodesic Packet Forwarding (AGPF), Anchored Path Discovery (APD), multipath routing and path maintenance. We performed simulations of the TLR and TRR protocols using the GloMoSim simulator. In order to do that, we have implemented a new mobility model that we call ``restricted random waypoint

The integration of voice, data, and video services modified the target of networking technologies. Now, instead of simply providing best effort delivery, the network has also to address integration of services and quality of service (QoS) support. The applications should be therefore provided with means to describe the service they want to receive from the network. In the Internet, two major mechanisms have been studied and developed: Integrated Services and Differentiated Services. Combining Integrated Services (R. Braden et al., 1994), QoS mechanisms for end-to-end signaling at hosts and Differentiated Services (S. Blake et al., 1998), QoS mechanisms in the network core is one proposal for overcoming the limitation of either mechanisms when applied alone (Y. Bernet et al., 1999). Some of the open issues of Y. Bernet et al. are discussed and a concrete proposal is presented, which maps the IntServ Controlled Load Service into the DiffServ Expedited Forwarding "Per Hop Behavior". The proposed approach has been implemented under the Linux operating system. The implementation allows verifying of the correct functional behavior and provides a platform where more complete performance tests are possible.

For pt. I see ibid., vol.4, p.93-6 (May 2000). We model some queuing systems arising in guaranteed service networks (such as RSVP/IP or ATM) as nonlinear min-plus systems that can be bounded by linear systems. We apply this method to the window flow control problem previously studied by Chang (1997), Agrawal and Rajan (1996), to the optimal smoothing of video through a network offering guaranteed service. We revisit the greedy shaper, and we also show how the same method enables us to compute the losses in a shaper by modelling it as a linear min-plus system. Finally, we describe the time-varying shaper.

Terminodes are personal devices that provide functionality of both the terminals and the nodes of the network. A network of terminodes is an autonomous, fully self-organized, wireless network, independent of any infrastructure. It must be able to scale up to millions of units, without any fixed backbone or server. In this paper we present the main challenges and discuss the main technical directions.

We consider the multiplexing of several variable bit rate (VBR) connections over one variable bit rate connection where the multiplexing uses a multiplexing buffer of size B. The VBR trunk is itself a connection and has a multidimensional connection descriptor, reflecting peak and sustainable rates. Given a cost function for the VBR trunk and a connection admission control (CAC) method for the input connections, we focus on the problem of finding the VBR trunk connection descriptor that minimizes the cost function and is able to accept set given set of VBR input connections. First, we show that, under reasonable assumptions on the cost function, the optimization problem can be reduced to a simpler one. Then we consider the homogeneous, loss-free case, for which we give an explicit CAC method. In this case, we find that, for all reasonable cost functions, the optimal VBR trunk is either of the CBR type, or is truly VBR, with a burst duration equal to the burst duration of the input connections. We show that the optimal peak cell rate is fixed for a given B (thus for a CBR trunk), and a VBR choice can only be an improvement. Lastly, we take as an example of the cost function the equivalent capacity of the VBR trunk. These results are expected to form the basis for a general method for a connection manager at a multiplexing node in an integrated services packet network.