Current radio frequency (RF) sensors at the Edge lack the computational resources to support practical, in-situ training for intelligent spectrum monitoring. This is true for sensor data classification in general. We propose a solution via Deep Delay Loop Reservoir Computing (DLR), a processing architecture that supports general machine learning algorithms on compact mobile devices by leveraging delay-loop reservoir computing in combination with innovative electro-optical hardware. With both digital and photonic realizations of our design of the loops, DLR delivers reductions in form factor, hardware complexity and latency, compared to the State-of-the-Art (SoA). The main impact of the reservoir is to project the input data into a higher dimensional space of reservoir state vectors in order to linearly separate the input classes. Once the classes are well separated, traditionally complex, power-hungry classification models are no longer needed for the learning process. Yet, even with simple classifiers based on Ridge regression (RR), the complexity grows at least quadratically with the input size. Hence, the hardware reduction required for training on compact devices is in contradiction with the large dimension of state vectors. DLR employs a RR-based classifier to exceed the SoA accuracy, while further reducing power consumption by leveraging the architecture of parallel (split) loops. We present DLR architectures composed of multiple smaller loops whose state vectors are linearly combined to create a lower dimensional input into Ridge regression. We demonstrate the advantages of using DLR for two distinct applications: RF Specific Emitter Identification (SEI) for IoT authentication, and wireless protocol recognition for IoT situational awareness.

We describe three in-field data collection efforts yielding a large database of RSSI values vs. time or distance from vehicles communicating with each other via DSRC. We show several data processing schemes we have devised to develop opportunistic Vehicle-to-Vehicle (V2V) propagation models from such data. The database is limited in several important ways, not least, the presence of a high noise floor that limits the distance over which good modeling is feasible. Another is the presence of interference from multiple active transmitters. Our methodology makes it possible to obtain, despite these limitations, accurate models of median path loss vs. distance, shadow fading, and fast fading caused by multipath. We aim not to develop a new V2V model, but to show the methods enabling such a model to be obtained from in-field RSSI data, without elaborate measurement design and the associated deployment cost. Finally, models based on field data allow for capturing the multiple effects of an increasing number of simultaneous V2V transceivers under typical extreme traffic scenarios.

We propose an efficient application layer coding scheme suitable for time-limited wireless broadcast framework of the MBMS standards. The scheme, referred to as doped broadcast, is based on Fountain codes, and uses feedback to implement and control the tradeoff between the reconstruction delay, broadcast overhead, and decoding time/complexity. As the standardized schemes, our doped broadcast operates in two phases consisting of the limited time broadcast followed by an individualized repair phase in order to ensure (possibly prioritized) quality of service (QoS) to most users. The goal of the scheme is not to improve on any particular performance metric where highly optimized standard recommendations already perform exceptionally well, but rather to enable flexible and transparent mechanisms to implement and control tradeoff between different performance metrics. Toward this goal, we develop an analytically tractable model for doped broadcast with Ideal Soliton based codes leading to a repair strategy parameter estimation. The impact that inactivation and doping mechanisms employed by the decoder have on the complexity and overhead metrics is quantified and discussed for the proposed model. Hence, our approach guides a practical design tradeoff, which is important in today's highly heterogeneous environments requiring individualized QoS.

We propose a compression/transmission scheme that allows the quality of the reconstructed signal to gracefully degrade as the channel quality drops, as well as steadily improve with the channel improvement. The main idea is to partition the channel and/or network resources into m units (e.g., sub-bands, packets) and compress the source independently m times to perfectly match single unit resources, thus creating m independently distorted source versions. Consequently, we create a multiple-description, joint source-channel like architecture, that enables efficient reconstruction starting from a single received description with improvements onward. We further split the compression rate in two parts, allocating one to a rate-distortion optimal encoder, and the other to transmitting uncoded source symbols. We show how this architecture can easily leverage modularity in terms of adjustable rate-splitting ratio and the maximum number of descriptions, e.g., through software parameters, to simultaneously and robustly (i.e. avoiding the cliff effect) achieve operating points close to rate-distortion curve for many channel states. We demonstrate how statistical description of channel states (or performance statistics of content delivery network) can be used to set the two parameters constructively in terms of converging to optimal operation in the range of interest.

This paper studies source location anonymity in a large monitoring wireless sensor network with a single data collector, and under a global attack. The qualifier ”global” indicates the capability of the eavesdropper (attacker) to capture all network transmissions, and to discern their time and location. We propose a scheme for generating fake network traffic to disguise the real event notification. This scheme is particularly effective for the protection of the monitored asset in delay-intolerant applications monitoring rare and spatially sparse events. Unlike earlier work on this topic, we jointly consider the protection strength, events' dynamics, probability of the attacker's exposure during the attack, notification latency, network overhead, and scalability. The efficiency of the scheme that provides statistical source anonymity is achieved by partitioning network nodes randomly into several node groups. Members of the same group collectively emulate both temporal and spatial distribution of the event. Under such framework, we aim to better model the global eavesdropper, especially her way of using statistical tests to detect the real event. In addition, our approach aims to reduce the per-event work spent to generate the fake traffic while, most importantly, providing a guaranteed latency in reporting the event. The latency is controlled by decoupling the routing from the fake-traffic schedule. A good dummy-source group design also provides a robust protection of event bursts. This is achieved at the expense of the significant overhead as the number of dummy-source groups must be increased to the reciprocal value of the false alarm parameter used in the statistical test. Ultimately, our message is that designing a protection scheme to meet multiple requirements, imposed by realistic application scenarios, involves trade-offs among several performance measures, and calls for an evaluation framework that recognizes these challenges. We believe that the proposed source anonymity protection strategy, and the evaluation framework, are well justified by the abundance of the applications that monitor a rare event with uniform spatial distribution.

This paper studies decentralized, Fountain and network-coding based strategies for facilitating data collection in circular wireless sensor networks, which rely on the stochastic diversity of data storage. The goal is to allow for a reduced delay collection by a data collector who accesses the network at a random position and random time. Data dissemination is performed by a set of relays which form a circular route to exchange source packets. The storage nodes within the transmission range of the route's relays linearly combine and store overheard relay transmissions using random decentralized strategies. An intelligent data collector first collects a minimum set of coded packets from a subset of storage nodes in its proximity, which might be sufficient for recovering the original packets and, by using a message-passing decoder, attempts recovering all original source packets from this set. Whenever the decoder stalls, the source packet which restarts decoding is polled/doped from its original source node. The random-walk-based analysis of the decoding/doping process furnishes the collection delay analysis with a prediction on the number of required doped packets. The number of doped packets can be surprisingly small when employed with an Ideal Soliton code degree distribution and, hence, the doping strategy may have the least collection delay when the density of source nodes is sufficiently large. Furthermore, we demonstrate that network coding makes dissemination more efficient at the expense of a larger collection delay. Not surprisingly, a circular network allows for a significantly more (analytically and otherwise) tractable strategies relative to a network whose model is a random geometric graph.

For wireless sensor networks with many locationunaware nodes, which can be modeled as a planar Poisson point process, we investigate a protocol, dubbed BeSpoken, which steers data transmissions along a straight path called a spoke. BeSpoken implements a simple, spatially recursive process, where a basic set of control packets and a data packet are exchanged repeatedly among daisy-chained relays that constitute the spoke. Hence, a data packet originated by the first relay makes a forward progress in the direction of the spoke. Despite the simplicity of the protocol engine, modeling the spoke process is a significant challenge. Bespoken directs data transmissions by randomly selecting relays to retransmit data packets from crescent-shaped areas along the spoke axis. The resulting random walk of the spoke hop sequence may be modeled as a two dimensional Markov process. Based on this model, we propose design rules for protocol parameters that minimize energy consumption while ensuring that spokes propagate far enough and have a limited wobble with respect to the spoke axis. The energy efficiency is demonstrated through simulations of the BeSpoken-based data search, and a comparison with the energy consumption of a search based on directed diffusion.