Conformal prediction yields a prediction set with guaranteed $1-\alpha$ coverage of the true target under the i.i.d. assumption, which can fail and lead to a gap between $1-\alpha$ and the actual coverage. Prior studies bound the gap using total variation distance, which cannot identify the gap changes under distribution shift at different $\alpha$, thus serving as a weak indicator of prediction set validity. Besides, existing methods are mostly limited to covariate shifts, while general joint distribution shifts are more common in practice but less researched. In response, we first propose a Wasserstein distance-based upper bound of the coverage gap and analyze the bound using probability measure pushforwards between the shifted joint data and conformal score distributions, enabling a separation of the effect of covariate and concept shifts over the coverage gap. We exploit the separation to design algorithms based on importance weighting and regularized representation learning (WR-CP) to reduce the Wasserstein bound with a finite-sample error bound. WR-CP achieves a controllable balance between conformal prediction accuracy and efficiency. Experiments on six datasets prove that WR-CP can reduce coverage gaps to 3.2% across different confidence levels and outputs prediction sets 37% smaller than the worst-case approach on average.

Differential privacy is a widely accepted measure of privacy in the context of deep learning algorithms, and achieving it relies on a noisy training approach known as differentially private stochastic gradient descent (DP-SGD). DP-SGD requires direct noise addition to every gradient in a dense neural network, the privacy is achieved at a significant utility cost. In this work, we present Spectral-DP, a new differentially private learning approach which combines gradient perturbation in the spectral domain with spectral filtering to achieve a desired privacy guarantee with a lower noise scale and thus better utility. We develop differentially private deep learning methods based on Spectral-DP for architectures that contain both convolution and fully connected layers. In particular, for fully connected layers, we combine a block-circulant based spatial restructuring with Spectral-DP to achieve better utility. Through comprehensive experiments, we study and provide guidelines to implement Spectral-DP deep learning on benchmark datasets. In comparison with state-of-the-art DP-SGD based approaches, Spectral-DP is shown to have uniformly better utility performance in both training from scratch and transfer learning settings.

Membership inference attacks (MIAs) against machine learning models lead to serious privacy risks for the training dataset used in the model training. The state-of-the-art defenses against MIAs often suffer from poor privacy-utility balance and defense generality, as well as high training or inference overhead. To overcome these limitations, in this paper, we propose a novel, lightweight and effective Neuron-Guided Defense method named NeuGuard against MIAs. Unlike existing solutions which either regularize all model parameters in training or noise model output per input in real-time inference, NeuGuard aims to wisely guide the model output of training set and testing set to have close distributions through a fine-grained neuron regularization. That is, restricting the activation of output neurons and inner neurons in each layer simultaneously by using our developed class-wise variance minimization and layer-wise balanced output control. We evaluate NeuGuard and compare it with state-of-the-art defenses against two neural network based MIAs, five strongest metric based MIAs including the newly proposed label-only MIA on three benchmark datasets. Extensive experimental results show that NeuGuard outperforms the state-of-the-art defenses by offering much improved utility-privacy trade-off, generality, and overhead. Our code is publicly available at https://github.com/nux219/NeuGuard.

In applications such as participatory sensing and crowd sensing, self-interested agents exert costly effort towards achieving an objective for the system operator. We study such a setup where a principal designs a sequence of contracts to incentivize multiple agents of different types to achieve a global objective. The agents can collude with each other to derive rent, since the principal cannot observe the efforts exerted directly, but only the outcome of the task which is a noisy function of the effort. The type of each agent influences the effort cost and task output. For a duopoly in which agent payments are coupled, we show that i) if the principal and the agents interact finitely many times, the agents can derive rent by colluding even if the principal knows the types of the agents, and ii) if the principal and the agents interact infinitely often, the principal can disincentivize agent collusion through a dynamic data-driven contract.

Data-driven anomaly detection over time series data is studied from the perspective of separating data anomalies—corresponding to sensor failures—from process anomalies—that arise from equipment or operational failures. A semi-supervised approach is proposed that utilizes two predictive models trained on non-anomalous data using two different sensor groups as inputs, and a nested hypothesis test to reliably classify data or process anomalies. Conditions are derived on choice of sensor groups to guarantee reliable detection, and a case study is presented to demonstrate the proposed classification approach.

This work presents a new paradigm for protection of sensitive inferences drawn from data streams with relevance to Internet-of-Things (IoT). This paradigm is an alternative to end-to-end encryption of entire data streams, or noise-addition based privatization mechanisms. It relies on the notion that raw data shared through IoTs are themselves not sensitive but for the inferences that can be drawn from them, and further, these inferences vary much slower than the collected data. Methodologies are developed that transform data streams into two parallel sub-streams of minimum sufficient and maximal irrelevant statistics, such that the sparse minimal sufficient stream can be protected using encryption, and the high rate irrelevant stream is guaranteed to provide perfect privacy for the underlying inference without any additional protection. This inferential separation is explored theoretically, where it is proved that the inference relevant (minimum sufficient) stream grows asO(logt) for a data stream of lengtht. The approach is extended to bandwidth constrained devices, where a new optimal quantization scheme is presented that achieves maximum fidelity while guaranteeing privacy. The presented algorithms are demonstrated to practical IoT datasets where trained CNN based classifiers are shown to fail on the unprotected high rate stream.

A user implemented privacy preservation mechanism is proposed for the online gradient descent (OGD) algorithm. Privacy is measured through the information leakage as quantified by the mutual information between the users outputs and learners inputs. The input perturbation mechanism proposed can be implemented by individual users with a space and time complexity that is independent of the horizon T. For the proposed mechanism, the information leakage is shown to be bounded by the Gaussian channel capacity in the full information setting. The regret bound of the privacy preserving learning mechanism is identical to the non private OGD with only differing in constant factors.

Cyber physical systems which integrate physical system dynamics with digital cyber infrastructure are envisioned to transform our core infrastructural frameworks, such as the smart electricity grid, transportation networks, and advanced manufacturing. This integration, however, exposes the physical system functioning to the security vulnerabilities of cyber communication. Both scientific studies and real-world examples have demonstrated the impact of data injection attacks on complex systems, including the Internet, the smart electricity grid, and air traffic systems. In this paper, an abstract theoretical framework is proposed to study data injection/modification attacks on Markov modeled dynamical systems from the perspective of an adversary. Typical data injection attacks focus on one shot attacks by adversary and the non-detectability of such attacks under static assumptions. In this paper, we study dynamic data injection attacks where the adversary is capable of modifying a temporal sequence of data and the physical controller is equipped with prior statistical knowledge about the data arrival process to detect the presence of an adversary. The goal of the adversary is to modify the arrivals to minimize a utility function of the controller while minimizing the detectability of his presence as measured by the K-L divergence between the prior and posterior distribution of the arriving data. The tradeoff between these two metrics-controller utility and the detectability cost-is studied analytically for different underlying dynamics. The proposed framework is then applied to a practical problem in data networks where a router tries to hide the path of traffic flow from timing analysis by an active adversary who can modify the timing of an incoming packet stream. This problem is studied from the adversary perspective wherein the goal is to balance two costs-the adversary's detectability cost measured by the K-L divergence and the network privacy cost measured by the maximum length of the packet stream whose paths can be hidden by a memory limited router.

The problem of false data injection through compromised cyber links to a physical control system modeled by linear quadratic Gaussian dynamics is studied in this paper. The control input stream is compromised by an attacker who modifies the (cyber) control signals transmitted with the objective of increasing the quadratic cost incurred by the (physical) controller whilst maintaining a degree of stealthiness. The tradeoff between the increase in quadratic cost and the stealthiness (or detectability), are measured by the Kullback-Leibler distance between legitimate and falsified state dynamics is characterized analytically. It is shown that the optimal adversarial strategy is a sequence of independent Gaussian noise signals with carefully chosen variances whose eigenvalues align with those of the legitimate noise covariance with the scaling reflecting the desired quadratic cost increase. As the stealthiness decreases, the optimal tradeoff is shown to be linear with slope inversely proportional to the maximal of maximal eigenvalue of modified reward matrices. Numerical simulations are presented that showcase the optimal tradeoff and the comparison of the legitimate and falsified dynamics under different requirements on detectability.

Anonymous systems on the Internet aim to protect users from revealing to an external unauthorized entity their identities and their network activities. Despite using layered encryption, these systems are still vulnerable to timing analysis, wherein an eavesdropper can use traffic correlation mechanisms to identify the source of packets arriving at a destination. Mixes are intelligent routers or proxy servers that aim to provide packet source anonymity from timing analysis by delaying and shuffling the order of received packets prior to transmission. Such shuffling strategies naturally increase latency and result in a tradeoff between anonymity and latency. This paper investigates this tradeoff in a network of mixes, by deriving the optimal routing for sources which maximizes weighted sum of anonymity and delay. The achievable anonymity is characterized analytically for a general multipath model, and it is shown that under light traffic conditions, there exists a unique single route strategy, which achieves the optimal delay anonymity tradeoff. A low complexity algorithm is presented that derives the optimal routes to achieve a desired tradeoff. The light traffic results are specialized for a graphical model of existing practical anonymous systems, and optimal scaling behavior with the size of such networks is characterized. In the heavy traffic regime, it is shown that optimal anonymity is achieved for any allocation of rates across the different routes. Simulations on example networks are presented where it is shown that the optimal routes derived under light traffic performs quite well in general traffic regime.

In this work, we study information leakage in timing side channels that arise in the context of shared event schedulers. Consider two processes, one of them an innocuous process (referred to as Alice) and the other a malicious one (referred to as Bob), using a common scheduler to process their jobs. There are other innocuous users in addition to Alice and Bob using the scheduler to process their jobs. Based on when his jobs get processed, Bob wishes to learn about the pattern (size and timing) of Alice's jobs. Depending on the context, knowledge of this pattern could have serious implications on Alice's privacy and security. For instance, shared routers can reveal traffic patterns, shared memory access can reveal cloud usage patterns, and suchlike. We present a formal framework to study the information leakage in shared resource schedulers using the pattern estimation error as a performance metric. The first-come-first-serve (FCFS) scheduling policy and time-division-multiple-access (TDMA) are identified as two extreme policies on the privacy metric, FCFS has the least, and TDMA has the highest. However, on performance-based metrics, such as throughput and delay, it is well known that FCFS significantly outperforms TDMA. We then derive two parameterized policies, accumulate and serve, and proportional TDMA, which take two different approaches to offer a tunable trade-off between privacy and performance.

Fairness among multiple users sharing a common resource is an important criterion in the design and evaluation of scheduling algorithms in networks. Anonymous networking, where sources of transmitted packets are undecipherable to an eavesdropper, requires packets arriving at routers from multiple sources to be randomly reordered prior to transmission, which works against the notion of temporal fairness in packet scheduling. Consequently, it is important to understand the relationship between temporal fairness and achievable anonymity. In this paper, this relationship is investigated for three fair scheduling paradigms: First-Come-First-Serve (FCFS), Fair Queuing, and the Proportional Method. Using an information-theoretic metric for anonymity and a common temporal fairness index that measures the degree of out-of-order transmissions, the anonymity achievable under these scheduling paradigms is characterized and their anonymity-fairness tradeoffs are compared. The FCFS and Fair Queuing algorithms have little inherent anonymity, and a significant improvement in anonymity is achieved by relaxing their respective fairness paradigms. The analysis of the relaxed FCFS criterion, in particular, is accomplished by modeling the problem as a stochastic control system that is solved using dynamic programming. The proportional method of scheduling, while unpopular in networks today, is shown to outperform the other fair scheduling algorithms when trading temporal fairness for anonymity, and is also proven to be asymptotically optimal as the buffer size of the scheduler is increased.

Infrastructural systems such as the electricity grid, healthcare, and transportation networks today rely increasingly on the joint functioning of networked information systems and physical components, in short, on cyber-physical architectures. Despite tremendous advances in cryptography, physical-layer security and authentication, information attacks, both passive such as eavesdropping, and active such as unauthorized data injection, continue to thwart the reliable functioning of networked systems. In systems with joint cyber-physical functionality, the ability of an adversary to monitor transmitted information or introduce false information can lead to sensitive user data being leaked or result in critical damages to the underlying physical system. This paper investigates two broad challenges in information security in cyber-physical systems (CPSs): preventing retrieval of internal physical system information through monitored external cyber flows, and limiting the modification of physical system functioning through compromised cyber flows. A rigorous analytical framework grounded on information-theoretic security is developed to study these challenges in a general stochastic control system abstraction-a theoretical building block for CPSs-with the objectives of quantifying the fundamental tradeoffs between information security and physical system performance, and through the process, designing provably secure controller policies. Recent results are presented that establish the theoretical basis for the framework, in addition to practical applications in timing analysis of anonymous systems, and demand response systems in a smart electricity grid.

Anonymous communication, where users communicate without revealing the identities of communicating parties or the paths of data flow is critical in data networks. On the Internet, Chaum mixes, intermediate nodes, or proxy servers, which use layered encryption and packet shuffling methods to hide source identities, are used to provide anonymity to network users. In this paper, an information theoretic framework is developed to study the maximum anonymity achievable by packet shuffling when the mixes are memory limited-in other words, they can store a finite number of packets. Using the Shannon entropy of the a posteriori distribution of packet sources from an eavesdropper's perspective as the measure of anonymity, the maximum achievable anonymity of a single mix with buffer size b (packets) serving two independent Poisson sources with equal arrival rates is shown to be log [2 cos (π/b+3)]. For a general multiuser b+3 system, the maximum anonymity as buffer size b → ∞ is shown to approach the entropy of the source arrival probabilities at a convergence rate no lesser than 1/b2. When the arrival probabilities of the general multiuser system can be expressed as a rational fraction k/2nfor some fixed n, this convergence rate is shown to be achievable. The anonymity analysis is extended to a general network of mixes connecting the sources to a common destination, where the source anonymity achievable on the destination link is shown to be lower bounded by a weighted sum of the anonymity achievable by each individual mix.

Cyber physical systems which integrate physical system dynamics with digital cyber infrastructure are envisioned to transform our core infrastructural frameworks such as the smart electricity grid, transportation networks and advanced manufacturing. This integration however exposes the physical system functioning to the security vulnerabilities of cyber communication. Both scientific studies and real world examples have demonstrated the impact of data injection attacks on state estimation mechanisms on the smart electricity grid. In this work, an abstract theoretical framework is proposed to study data injection/modification attacks on Markov modeled dynamical systems from the perspective of an adversary. Typical data injection attacks focus on one shot attacks by adversary and the non-detectability of such attacks under static assumptions. In this work we study dynamic data injection attacks where the adversary is capable of modifying a temporal sequence of data and the physical controller is equipped with prior statistical knowledge about the data arrival process to detect the presence of an adversary. The goal of the adversary is to modify the arrivals to minimize a utility function of the controller while minimizing the detectability of his presence as measured by the KL divergence between the prior and posterior distribution of the arriving data. Adversarial policies and tradeoffs between utility and detectability are characterized analytically using linearly solvable control optimization.

Variable Bitrate Coding (VBR) has shown to be an advantageous method of encoding data streams, with particular application to speech, audio and video streams. While the primary disadvantage of VBR has long been considered as the increasing encoding complexity, recent research into traffic analysis of VBR coded audio streams has exposed an important privacy vulnerability wherein an eavesdropper can utilize the observed length of VBR encoded data packets and determine the contents of the communication such as spoken words and audio. In this work, the privacy-utility tradeoff for VBR coding is studied from a theoretical foundational perspective. Specifically, the data source is modeled a mixture distribution, wherein the length of the encoded data packet is varied to maintain a constant quality (distortion) with respect to the source. Using Shannon's equivocation as a measure of data privacy, the tradeoff between privacy and utility (as measured by delay and overall bit rate) of VBR is investigated analytically. In particular, the tradeoffs are expressed as classical information theoretic rate distortion functions, which shed light into methods to increase the privacy of VBR encoded data without compromising on the desired output fidelity.

As networked systems increasingly pervade every facet of life, it is quintessential for users to communicate without revealing their identities or the paths of data flow. Chaum Mixes are intermediate nodes or routers that are used to provide anonymity by using cryptographic and batching methods to hide source identities. The anonymity achievable by batching strategies, are however, severely impacted by limited buffer capacity of the mix node. This paper presents an information theoretic investigation of a buffer constrained mix, and provides the first single letter characterization of the maximum achievable anonymity as a function of buffer size for a mix serving two users with equal arrival rates. For two users with unequal arrival rates the anonymity is expressed as a solution to a series of finite recursive equations. For more than two users and arbitrary arrival rates, a lower bound on the convergence rate of anonymity is derived as buffer size increases and it is shown that under certain arrival configurations the lower bound is tight.

Timing analysis has long been used to compromise user anonymity in networks. Even when data is encrypted, an adversary can track flows from sources to the corresponding destinations by merely using the correlation between the inter packet timing on incoming and outgoing streams at intermediate routers. Anonymous network systems, where users communicate without revealing their identities, rely on the idea of Chaum mixing to hide `networking information'. Chaum mixes are routers or proxy servers that randomly reorder the outgoing packets to prevent an eavesdropper from tracking the flow of packets. The effectiveness of such mixing strategies is, however, diminished under constraints on network resources such as memory and bandwidth. In this work, a detection theoretic framework is proposed to study the optimization of mixing strategies under such constraints. Specifically, using the detection time of the adversary as a metric, the effectiveness of mixing strategies is maximized under constraints on memory and throughput. A general game theoretic model is proposed to study the mixing strategies when an adversary is capable of capturing a fraction of incoming packets. For the proposed multistage game, existence of a Nash equilibrium is proven, and the optimal strategies for the mix and adversary are derived at the equilibrium condition.

Fairness amongst multiple users sharing a common resource has been an important criterion in the evaluation of scheduling algorithms in networks. Anonymous networking, where sources of transmitted packets are undecipherable to an eavesdropper, requires that packets from multiple sources are randomly reordered prior to transmission which works against the notion of fair scheduling. Consequently, it is important to understand the relationship between fairness and achievable anonymity in networking. In this paper, this relationship is characterized for the class of fair scheduling axioms defined by considering the equal treatment ex ante and demand mono-tonicity, under which the proportional method is known to be the unique scheduling algorithm that achieves the desired fairness. Using an information theoretic quantitative framework, the anonymity of this scheduling algorithm is characterized and proven to be asymptotically optimal with increase in buffer size. The anonymity achieved by the proportional method is also shown to be significantly better than conventional fair scheduling algorithms such as first come first serve and round robin, thus making a case for its application in data networks.

In this work, we study information leakage in timing side channels that arise in the context of shared event schedulers. Consider two processes, one of them an innocuous process (referred to as Alice) and the other a malicious one (referred to as Bob), using a common scheduler to process their jobs. Based on when his jobs get processed, Bob wishes to learn about the pattern (size and timing) of jobs of Alice. Depending on the context, knowledge of this pattern could have serious implications on Alice's privacy and security. For instance, shared routers can reveal traffic patterns, shared memory access can reveal cloud usage patterns, and suchlike. We present a formal framework to study the information leakage in shared resource schedulers using the pattern estimation error as a performance metric. In this framework, a uniform upper bound is derived to benchmark different scheduling policies. The first-come-first-serve scheduling policy is analyzed, and shown to leak significant information when the scheduler is loaded heavily. To mitigate the timing information leakage, we propose an “Accumulate-and-Serve” policy which trades in privacy for a higher delay. The policy is analyzed under the proposed framework and is shown to leak minimum information to the attacker, and is shown to have comparatively lower delay than a fixed scheduler that preemptively assigns service times irrespective of traffic patterns.

In multi-tasking systems where a finite resource is to be shared, a scheduler dictates how the resource is divided among competing processes. Examples of systems which have schedulers include, a computer where the CPU needs to be shared between the different threads running, a cloud computing infrastructure with shared computing resources, a network router serving packets from different streams etc. In such situations, when a processor is shared by multiple users, the delays experienced by jobs from one user are a function of the arrival pattern of jobs from other users, and the scheduling policy of the server. Consequently, a scheduling system creates a timing side channel in which information about arrival pattern from one user is inadvertently leaked to another. In this work, this information leakage is studied for a two user scheduling system. We first introduce a measure of privacy and then demonstrate that no scheduler can provide maximum privacy without idling/taking vacations, and consequently no policy can simultaneously be delay and privacy optimal.

Anonymous wireless networking is studied when an adversary monitors the transmission timing of an unknown subset of the network nodes. For a desired quality of service (QoS), as measured by network throughput, the problem of maximizing anonymity is investigated from a game-theoretic perspective. Quantifying anonymity using conditional entropy of the routes given the adversary's observation, the problem of optimizing anonymity is posed as a two-player zero-sum game between the network designer and the adversary: The task of the adversary is to choose a subset of nodes to monitor so that anonymity of routes is minimum, whereas the task of the network designer is to maximize anonymity by choosing a subset of nodes to evade flow detection by generating independent transmission schedules. In this two-player game, it is shown that a unique saddle-point equilibrium exists for a general category of finite networks. At the saddle point, the strategy of the network designer is to ensure that any subset of nodes monitored by the adversary reveals an identical amount of information about the routes. For a specific class of parallel relay networks, the theory is applied to study the optimal performance tradeoffs and equilibrium strategies. In particular, when the nodes employ transmitter-directed signaling, the tradeoff between throughput and anonymity is characterized analytically as a function of the network parameters and the fraction of nodes monitored. The results are applied to study the relationships between anonymity, the fraction of monitored relays, and the fraction of hidden relays in large networks.

The information leakage of a queuing side channel in two-user-shared scheduling system is studied from an information theoretic perspective. In the queueing side channel, a malicious attacker can learn the pattern of jobs from a legitimate user using the queuing delays experienced at the shared buffer. An analytical framework is proposed to quantify information leakage using Shannon's equivocation, and the information leakage of the standard First-come-First-serve scheduler is studied in a slotted system with geometric arrivals. The analysis of the FCFS scheduler demonstrates that the policy provides “good privacy” when arrival rates are very low; the leaked information increases with the rate of the attacker's jobs and approaches the maximum retrievable information as the sum-rate of arrivals approaches the boundary of the stability region of the queue.

No abstract available.

We examine a queuing side channel which results from a shared resource between two users in the context of packet networks. We consider the scenario where one of them is a legitimate user and the other is an attacker who is trying to learn about the former's activities. We show that the waiting time of an adversary sending a small but frequent probe stream to the shared resource (e.g., a router) is highly correlated with traffic pattern of the user.

Chaum Mixes, a class of proxy servers or relays which use randomized reordering and batching of packets from multiple users to provide source anonymity, have been used extensively for anonymous remailing, browsing and peer-to-peer file sharing. In this work, an analytical framework is proposed to measure and optimize the anonymity provided by mixing strategies when the mixes have memory restrictions. Specifically, an information-theoretic metric of anonymity is proposed for buffer constrained mixes, and using Poisson traffic models, fundamental trade-offs between achievable anonymity and the buffer size are studied analytically. In particular, a buffer-constrained mixing strategy is proposed that is asymptotically optimal and obtains the best convergence rate amongst known mixing strategies. The strategy is generalized to a network of mixes, where the achievable anonymity is expressed as a function of the topology and the buffer constraints of individual mixes.

Providing anonymity to routes in a wireless ad hoc network from passive eavesdroppers is considered. Using Shannon's equivocation as an information theoretic measure of anonymity, scheduling strategies are designed for wireless nodes using receiver directed signaling. The achievable rate region for multiaccess relays are characterized under constraints on average packet latency. The relationship between overall network throughput and the route anonymity is obtained by drawing a connection to the rate-distortion tradeoff in information theory. A decentralized implementation of the relaying strategy is proposed, and the corresponding performance analyzed.

The information-theoretic analysis of Chaum mixing under latency constraints is considered. Mixes are relay nodes that collect packets from multiple users and modify packet timings to prevent an eavesdropper from identifying the sources of outgoing packets. In this work, an entropy-based metric of anonymity is proposed to quantify the performance of a mixing strategy under strict delay constraints. Inner and outer bounds on the maximum achievable anonymity are characterized as functions of traffic load and the delay constraint. The bounds are shown to have identical first derivatives at low traffic loads.

The problem of security against timing based traffic analysis in multihop networks is considered in this work. In particular, the relationship between the level of anonymity provided and the quality of service, as measured by network latency, is analyzed theoretically. Using an information theoretic measure of anonymity of routes in eavesdropped networks is considered, and packet scheduling strategies are designed to guarantee any desired level of anonymity. In particular, for individual relays, scheduling strategies based on mixing are designed so that the incoming and outgoing transmission epochs do not reveal any information. The proposed strategies utilize a limited fraction of dummy transmissions, and a significant reduction in packet latency at individual relays is demonstrated analytically for Poisson distributed arrivals. To minimize overall network latency, a randomized selection strategy is considered to choose the set of relays that use the designed scheduling strategies. The random selection is optimized for the desired level of anonymity using a well known distortion rate optimization in information theory. The tradeoff between overall network latency and anonymity in the network is characterized for centralized and decentralized scheduling strategies.

The problem of security against packet timing based traffic analysis in wireless networks is considered in this work. An analytical measure of ldquoanonymityrdquo of routes in eavesdropped networks is proposed using the information-theoretic equivocation. For a physical layer with orthogonal transmitter directed signaling, scheduling and relaying techniques are designed to maximize achievable network performance for any desired level of anonymity. The network performance is measured by the total rate of packets delivered from the sources to destinations under strict latency and medium access constraints. In particular, analytical results are presented for two scenarios: For a single relay that forwards packets from users, relaying strategies are provided that minimize the packet drops when the source nodes and the relay generate independent transmission schedules. A relay using such an independent scheduling strategy is undetectable by an eavesdropper and is referred to as a covert relay. Achievable rate regions are characterized under strict and average delay constraints on the traffic, when schedules are independent Poisson processes. For a multihop network with an arbitrary anonymity requirement, the problem of maximizing the sum-rate of flows (network throughput) is considered. A randomized selection strategy to choose covert relays as a function of the routes is designed for this purpose. Using the analytical results for a single covert relay, the strategy is optimized to obtain the maximum achievable throughput as a function of the desired level of anonymity. In particular, the throughput-anonymity relation for the proposed strategy is shown to be equivalent to an information-theoretic rate-distortion function.

We consider the design of quantizers for the distributed estimation of a deterministic parameter, when the fusion center uses a Maximum-Likelihood estimator. We define a new metric of performance, which is to minimize the maximum ratio between the Fisher Information of the unquantized and quantized observations. Since the estimator is M-L, the criterion is equivalent to the minimizing the maximum asymptotic relative efficiency due to quantization. We propose an algorithm to obtain the quantizer that optimizes the metric and prove its convergence. Through simulations, we illustrate that the quantizer performance is close to the best possible Fisher Information as number of quantization bits increases. Furthermore, under certain conditions, the quantizer structure is found to belong to the class of score-function quantizers, hich maximize Fisher Information for a given value of the parameter.

We consider random access and coding schemes for sensor networks with mobile access (SENMA). Using an orthogonal code-division multiple access (CDMA) as the physical layer, an opportunistic ALOHA (O-ALOHA) protocol that utilizes channel state information is proposed. Under the packet capture model and using the asymptotic throughput as the performance metric, we show that O-ALOHA approaches the throughput equal to the spreading gain with an arbitrarily small power at each sensor. This result implies that O-ALOHA is close to the optimal centralized scheduling scheme for the orthogonal CDMA networks. When side information such as location is available, the transmission control is modified to incorporate either the distribution or the actual realization of the side information. Convergence of the throughput with respect to the size of the network is analyzed. For networks allowing sensor collaborations, we combine coding with random access by proposing two coded random access schemes: spreading code dependent and independent transmissions. In the low rate regime, the spreading code independent transmission has a larger random coding exponent (therefore, faster decay of error probability) than that of the spreading code dependent transmission. On the other hand, the spreading code dependent transmission gives higher achievable rate.