Previous research on the ethical decision-making capabilities of large language models (LLMs) has largely relied on classification-based approaches, focusing on binary judgments of moral acceptability. While prior findings indicate strong alignment between LLMs and human judgments in morally unambiguous scenarios, the reasoning processes behind these decisions remain underexplored. This paper introduces an explainability-driven workflow to assess LLM moral reasoning by prompting models to provide both binary ethical judgments and accompanying justifications. We evaluate the semantic similarity of these justifications across varied prompting styles using latent semantic analysis. Gemini achieved the most consistent moral justification similarity across frameworks, with LLaMA3 close behind and LLaMA2 and Mistral showing greater variability, especially in deontology.

A new adaptive transmission protocol is introduced to improve the performance of slotted ALOHA. Nodes use known periodic schedules as base policies with which they collaboratively learn how to transmit periodically in different time slots so that packet collisions are minimized. The Adaptive Policy Tree (APT) algorithm is introduced for this purpose, which results in APT-ALOHA. APT-ALOHA does not require the presence of a central repeater and uses explicit acknowledgements to confirm the reception of packets. It is shown that nodes using APT-ALOHA quickly converge to transmission schedules that are virtually collision-free, and that the throughput of APT-ALOHA resembles that of TDMA, where slots are pre-allocated to nodes. In particular, APT-ALOHA attains a successful utilization of time slots- over 70% on saturation mode.

Decoy routing is a powerful circumvention mechanism intended to provide secure communications that cannot be monitored, detected, or disrupted by a third party who controls the user's network infrastructure. Current decoy routing protocols have weaknesses, however: they either make the unrealistic assumption that routes through the network are symmetric (i.e., the router implementing the decoy routing protocol must see all of the traffic, in both directions, from each connection it modifies), or their protocol requires modifying the route taken by packets in connections that use the protocol, and these route changes are detectable by a third party. We present Rebound, a decoy routing protocol that tolerates asymmetric routes without modifying the route taken by any packet that passes through the decoy router, making it more difficult to detect or disrupt than previous decoy routing protocols.

We address the problem of privacy in publish-subscribe(pub-sub) systems that typically expose some form of published content and subscriber interest, at least to the infrastructure responsible for subscription matching and content delivery. In our recent work, we proposed P3S, a pub-sub middleware designed to protect the privacy of subscriber interest and confidentiality of published content. P3S combined Cipher text Policy Attribute Based Encryption(CP-ABE) with Predicate Based Encryption (PBE) in its novel system architecture to achieve the desired level of content (payload and metadata) confidentiality, and subscription privacy. In this work, we build upon P3S to achieve the strongest possible subscription privacy where clear text subscription is visible only to the subscriber. Furthermore, we add support for subscription policy enforcement, improve the expressiveness of predicates by allowing disjunctions of conjunction, and improve the efficiency of the underlying cryptography through enhanced cryptographic construction and optimized implementation of cryptographic primitives. To the best of our knowledge, this paper presents the first comprehensive and practical implementation of a real-time privacy preserving pub-sub system, demonstrated on a large-scale test bed featuring up to 90 subscribers with robust, scalable and efficient performance. Our code and test bed specifications are freely available for research and experimentation purposes.

Spread spectrum (SS) communication relies on the assumption that some secret is shared beforehand among communicating nodes to establish the spreading sequence for long-term wireless communication. Strasser et al. identified this as the circular dependency problem (CDP). This problem is exacerbated in large networks, where nodes join and leave the network frequently, and preconfiguration of secrets through physical contact is infeasible. In this work, we introduce an efficient and adversary-resilient secret sharing mechanism based on two novel paradigms (intractable forward decoding, efficient backward decoding) called Time Reversed Message Extraction and Key Scheduling (TREKS) that enables SS communication without preshared secrets. TREKS is four orders of magnitude faster than previous solutions to the CDP. Furthermore, our approach can be used to operate long-term SS communication without establishing any keys. The energy cost under TREKS is provably optimal with minimal storage overhead, and computation cost at most twice that of traditional SS. We evaluate TREKS through simulation and empirically using an experimental testbed consisting of USRP, GNU Radio, and GPU-equipped nodes. Using TREKS under a modest hardware setup, we can sustain a 1--Mbps long-term SS communication spread by a factor of 100 (i.e., 100 Megachips per second) over a 200-MHz bandwidth in real time.

We study the problem of designing an efficient broadcast communication system in the presence of inside attackers from a Game Theory perspective. The network consists of a server that pre-assigns key sets to users some of whom may be malicious, either compromised or inherently selfish. The keys are mapped to communication channels unique in time and frequency, and the server broadcasts access information on those channels selectively. The goal of the server would be to deliver information to users using as fewer channels as possible. The malicious users, on the other hand, collude to selectively jam channels that their keys map to and keep as many users as possible from accessing the information while remaining undetected. This interaction of a server and inside attackers can be modeled as a 2-person game with server's payoff defined as the number of honest users who receive broadcast information as a function of the channels being used for broadcast, and the traitors' payoff defined as the number of honest users affected by jamming as a function of the channels being jammed. Our main contribution is the modeling of this server/intruder interaction as a non-cooperative game. Understanding that transmission as well as jamming cost depends on the key (channel) based on the number of users who own them and thus the amount of information it may reveal about a jammer, we assign a generic weight on the player's strategies yielding a generic payoff function. For simplicity, we first assume that both of the players are rational and that they have complete information about each other's payoff function and strategy space. With this, we derive a closed expression for the Nash Equilibrium (NE). Furthermore, we show that the complexity for computing the NE under our formulation is polynomial in the number of users. Then, we discuss preliminary steps required to arrive at the same game formulation without assuming complete information about player strategies, their payoffs and their rationality.

We investigate the resiliency of IEEE802.11 rate adaptation algorithms (RAA) against smart jamming attacks. We consider several classes of state-of-the-art RAAs that include the SampleRate, ONOE, AMRR, and the RAA used in Atheros Microsoft Windows XP driver. We model the behavior of these algorithms, and show the existence of very efficient attacks that exploit RAA-specific vulnerabilities as well as the inherent weaknesses that exist in the design of IEEE802.11 MAC and link layer protocol: in particular the overt packet rate information being transmitted, predictable rate selection mechanism, performance anomaly caused by the equiprobability of transmissions among all nodes regardless of the data rates being employed, and the lack of interference differentiation from poor link quality by IEEE802.11 RAAs. In this work, we present algorithms that determine optimal jamming strategies against RAAs for a given jamming budget, and experimentally demonstrate the efficiency of these smart jamming attacks, which can be orders of magnitude more efficient than naive jamming. For example, in the case of SampleRate, eight reactive jamming pulses every second are sufficient to achieve the same network throughput degradation achieved by a periodic jammer with the jamming energy cost 100 times higher. Some of the RAAs react even worse to smart jamming attacks; ONOE in particular suffers from the phenomenon of congestion collapse where the nodes fail to recover from the lowest data rate even after the jammer stops jamming. At the end, we summarize fundamental reasons behind such RAA vulnerabilities and propose a preliminary set of mitigation techniques. We leave the experimental demonstration of the efficiency of the proposed mitigation mechanisms for future work.

We consider the problem of key establishment over a wireless radio channel in the presence of a communication jammer, initially introduced in [13]. The communicating nodes are not assumed to pre-share any secret. The established key can later be used by a conventional spread-spectrum communication system. We introduce new communication concepts called intractable forward-decoding and efficient backward-decoding. Decoding under our mechanism requires at most twice the computation cost of the conventional SS decoding and one packet worth of signal storage. We introduce techniques that apply a key schedule to packet spreading and develop a provably optimal key schedule to minimize the bit-despreading cost. We also use efficient FFT-based algorithms for packet detection. We evaluate our techniques and show that they are efficient both in terms of resiliency against jammers and computation. Finally, our technique has additional features such as the inability to detect packet transmission until the last few bits are being transmitted, and transmissions being destination-specific. To the best of our knowledge, this is the first solution that is optimal in terms of communication energy cost with very little storage and computation overhead.

In this paper, we study the performance of the IEEE 802.11 MAC protocol under a range of jammers that covers both channel-oblivious and channel-aware jamming. We study two channel-oblivious jammers: a periodic jammer that jams deterministically at a specified rate, and a memoryless jammer whose signals arrive according to a Poisson process. We also develop new models for channel-aware jamming, including a reactive jammer that only jams non-colliding transmissions and an omniscient jammer that optimally adjusts its strategy according to current states of the participating nodes. Our study comprises of a theoretical analysis of the saturation throughput of 802.11 under jamming, an extensive simulation study, and a testbed to conduct real world experimentation of jamming IEEE 802.11 using GNU Radio and USRP platform. In our theoretical analysis, we use a discrete-time Markov chain analysis to derive formulae for the saturation throughput of IEEE 802.11 under memoryless, reactive and omniscient jamming. One of our key results is a characterization of optimal omniscient jamming that establishes a lower bound on the saturation throughput of 802.11 under arbitrary jammer attacks. We validate the theoretical analysis by means of Qualnet simulations. Finally, we measure the real-world performance of periodic and memoryless jammers using our GNU radio jammer prototype.

In this paper, we address the problem of countering jamming of broadcast control channels in wireless communication systems. Targeting control traffic on a system, e.g., BCCH channel in GSM, leads to smart attacks that can be four orders of magnitude more efficient than blind jamming. We propose several schemes based on coding theory and its applications that can counter both external and internal attackers (traitors). We introduce a T-(traitor) resilient scheme that requires less than (TlogTN)2control information transmissions and guarantees delivery of control information against any coalition of T traitors. The proposed scheme also allows the identification of persistently jamming traitors.