In this paper, we consider the problem of quickly detecting an abrupt change of linear coefficients in linear regression models. In particular, the observer sequentially observes a sequence of observations {(xn, yn)}∞n=1, which is assumed to obey a linear regression model at each time slot n. Some of the coefficients in the linear model change at a fixed but unknown time t. The post-change linear coefficients are unknown to the observer. The observer aims to design an online algorithm to detect the model change based on his sequential observations. Two performance metrics, namely the worst case detection delay (WADD) and the average run length to false alarm (ARL2FA), are adopted to evaluate the performance of detection algorithms. We design a low complexity algorithm, termed as parallel sum algorithm, for the detection purpose. An asymptotic upper bound on WADD is provided under any given ARL2FA constraint.

This letter considers localization using multiple-input multiple-output (MIMO) systems, configured with multiple transmit and receive sensors, widely distributed in a three-dimensional space. The placement of antennas is explored when the receiver hardware has varying noise quality. Cramer-Rao Lower Bounds are optimized to find the antenna placements, where it is shown that a symmetric configuration of transmitting and different quality receiving sensors around an emitter is optimal.

Exploiting physical layer characteristics to enhance or complement authentication strength in wireless networks has been attracting research attention recently. Existing physical layer authentication mechanisms mainly tackle single-hop communications. In this paper, we propose two physical layer challenge-response authentication mechanisms for wireless networks with relay. One mechanism, named PHY-CRAMR, is an extension of the existing PHY-CRAM protocol. It fully utilizes the randomness, reciprocity, and location decorrelation features of the wireless fading channel to hide/encrypt the challenge response messages at the physical layer, and is immune to outside attacks with a trusted relay. The other novel mechanism, named PHY-AUR, exploits randomness, coherence, and location decorrelation properties of wireless fading channel to securely convey the product of the channel state information on consecutive links and uses the fading channel to encrypt challenge and response messages. PHY-AUR is immune to both outside and inside attacks with an untrusted relay. Both PHY-CRAMR and PHY-AUR adopt OFDM technique to modulate the authentication key and challenge-response messages on subcarriers. Physical layer pilots and preambles are eliminated to prevent an attacker from gaining knowledge about the channel state information, and as a result prevent the authentication key from being revealed to untrusted attackers. We analyze the security strength of both mechanisms and conduct extensive simulations to evaluate them. It shows that both PHY-CRAMR and PHY-AUR can achieve both a high successful authentication rate and low false acceptance rate, and the performance improves as the signal to noise ratio (SNR) increases.

Most of the existing work on key generation from wireless fading channels requires a direct wireless link between legitimate users so that they can obtain correlated observations from the common wireless link. This paper studies the key generation problem in the two-way relay channel, in which there is no direct channel between the key generating terminals. We propose an effective key generation scheme that achieves a substantially larger key rate than that of a direct channel mimic approach. Unlike existing schemes, there is no need for the key generating terminals to obtain correlated observations in our scheme. We also investigate the effects of an active attacker on the proposed key generation protocol. We characterize the optimal attacker's strategy that minimizes the key rate of the proposed scheme. Furthermore, we establish the maximal attacker's power under which our scheme can still achieve a nonzero key rate.

The problem of simultaneously establishing multiple keys, one for each user in a set of users, is considered with possible assist from a group of dedicated helpers. For the case in which all users are required to generate keys, we develop a scheme that is sum rate optimal. For the case with dedicated helpers, we develop an achievable scheme and derive an outer bound. We identify conditions under which the developed scheme achieves the full capacity region and conditions under which it is sum rate optimal. We then specialize the study to a pairwise independent network model, for which we convert the key generation problem to a single-source multi-commodity flow over a network problem. Coupling results from graph theory, we fully characterize the capacity region for the general case of generating multiple keys with multiple helpers under the PIN model.

The paper considers the problem of distributed estimation of an unknown deterministic scalar parameter (the target signal) in wireless sensor networks (WSNs), in which each sensor receives a single snapshot of the field. The observation or sensing mode is only partially known at the corresponding nodes, perhaps, due to their limited sensing capabilities or other unpredictable physical factors. Specifically, it is assumed that the observation process at a node switches stochastically between two modes, with mode one corresponding to the desired signal plus noise observation mode (a valid observation), and mode two corresponding to pure noise with no signal information (an invalid observation). With no prior information on the local sensing modes (valid or invalid), the paper introduces a learning-based distributed estimation procedure, the mixed detection-estimation (MDE) algorithm, based on closed-loop interactions between the iterative distributed mode learning and estimation. The online learning (or sensing mode detection) step re-assesses the validity of the local observations at each iteration, thus refining the ongoing estimation update process. The convergence of the MDE algorithm is established analytically. Simulation studies show that, in the high signal-to-noise ratio (SNR) regime, the MDE estimation error converges to that of an ideal (centralized) estimator with perfect information about the node sensing modes. This is in contrast with the estimation performance of a naive average consensus based distributed estimator (with no mode learning), whose estimation error blows up with an increasing SNR.

Algorithms are studied for distributed least-squares (DLS) estimation of a scalar target signal in sensor networks. Due to the observation locality and the limited sensing ability, the individual sensor estimates are far from being reliable. To obtain a more reliable estimate of the target signal, the sensors could collaborate by iteratively exchanging messages with their neighbors, to refine their local estimates over time. Such an iterative DLS algorithm is investigated in this paper with and without the consideration of node failures. In particular, without sensor node failures it is shown that every instantiation of the DLS algorithm converges, i.e., consensus is reached among the sensors, with the limiting agreement value being the centralized least-squares estimate. With node failures during the iterative exchange process, the convergence of the DLS algorithm is still guaranteed; however, an error exists between the limiting agreement value and the centralized least-squares estimate. In order to reduce this error, a modified DLS scheme, the M-DLS, is provided. The M-DLS algorithm involves an additional weight compensation step, in which a sensor performs a one-time weight compensation procedure whenever it detects the failure of a neighbor. Through analytical arguments and simulations, it is shown that the M-DLS algorithm leads to a smaller error than the DLS algorithm, where the magnitude of the improvement dependents on the network topology.

This paper addresses the joint estimator and power optimization problem for a sensor network whose mission is to estimate an unknown parameter. We assume a two-hop network where each sensor collects observations from the source that transmits the quantity to be estimated, then amplifies and forwards its observations to a fusion center. The fusion center combines the observations using a Linear Minimum Mean Squared Error (LMMSE) estimator. We study the scenario where multiple parallel channels are available between the source and each sensor as well as between the sensors and the fusion center. We find the global optimal power allocation and estimator design for this network model. We present two practical scenarios of interest that utilize spatial and temporal diversity for which this solution applies, namely, a clustered network model and a single cluster model with an ergodic fading channel.