We develop a novel architecture called "Multi-Resolution Inter-Level Refinement (MR-ILR)" architecture for anomaly prediction in Internet of Things (IoT) data. Our architecture is comprised of three modules: First, the Inter-Level OR module takes the logical OR of the vector that represents the past anomaly states in IoT data and represents the occurrence of an anomaly state at increasingly coarse resolutions. Second, a Multi-Layer Perceptron (MLP) predicts the occurrence of anomaly states at any given resolution. Third, the anomaly predictions at successive resolutions are combined in Refiner modules to produce more accurate anomaly predictions. Our architecture provides the flexibility to produce anomaly predictions at distinct temporal resolutions. We compare the performance of our MR-ILR architecture against MLP and Long Short-Term Memory (LSTM) benchmark models. The results show that our architecture significantly outperforms both of these benchmark models with respect to the F1-score. This work represents an important advance in solving the challenging problem of anomaly prediction in IoT data and has the potential to be applied to a much wider range of problems that target anomaly prediction.

Blockage of Visible Light Communication (VLC) links by mobile obstacles such as humans is one of the key problems to solve in order to achieve the wide-scale deployment of indoor VLC networks. In this work, we present a solution to this problem by developing a predictive vertical handoff algorithm between Light Fidelity (Li-Fi) and Wireless Fidelity (Wi-Fi) networks. By using a state-of-the-art machine learning based forecasting model, our handoff algorithm predicts the number of time intervals for which blockage will occur in the next time block. Based on this prediction, our algorithm proactively hands off from Li-Fi to Wi-Fi and from Wi-Fi to Li-Fi in a manner that trades off the Average Available Data Rate (AADR) and the percentage service interruption. We demonstrate the performance of our algorithm on data collected in a life simulation environment in which humans move about in an indoor setting and block Li-Fi links. We show that our algorithm maintains a high AADR while achieving a very low percentage service interruption. Furthermore, we show that by varying the values of the parameters of our algorithm, we can achieve a gradual trade-off between AADR and the percentage service interruption. Our algorithm paves the way to high-performance hybrid Li-Fi/Wi-Fi networks that bear the potential to significantly change the landscape of indoor communication in the near future.

The massive access problem of the Internet of Things (IoT) is the problem of enabling the wireless access of a massive number of IoT devices to the wired infrastructure. In this article, we describe a multiscale algorithm (MSA) for joint forecasting-scheduling at a dedicated IoT gateway to solve the massive access problem at the medium access control (MAC) layer. Our algorithm operates at multiple time scales that are determined by the delay constraints of IoT applications as well as the minimum traffic generation periods of IoT devices. In contrast with the current approaches to the massive access problem that assume random arrivals for IoT data, our algorithm forecasts the upcoming traffic of IoT devices using a multilayer perceptron architecture and preallocates the uplink wireless channel based on these forecasts. The multiscale nature of our algorithm ensures scalable time and space complexity to support up to 6650 IoT devices in our simulations. We compare the throughput and energy consumption of MSA with those of reservation-based access barring (RAB), priority based on average load (PAL), and enhanced predictive version burst-oriented (E-PRV-BO) protocols, and show that MSA significantly outperforms these beyond 3000 devices. Furthermore, we show that the percentage control overhead of MSA remains less than 1.5%. Our results pave the way to building scalable joint forecasting-scheduling engines to handle a massive number of IoT devices at IoT gateways.

We formulate the automated MAC protocol generation problem under dynamic traffic conditions for multiple neighborhoods and in the presence of acknowledgments. We show that the problem can be formulated as a functional optimization program in which each design (a.k.a. decision) function of the program is the probability that a node takes an action given its knowledge state, as a function of the effective traffic demand at the current time at that node. In order to achieve a viable computational complexity for the functional optimization program, we discretize the effective traffic demands by virtue of which a look-up table is produced for each design function. Structurally different MAC protocols are subsumed in this framework, and are generated automatically with respect to traffic demand. The symbolic Monte Carlo method is used to generate an approximate expression for the objective function as well as for the non-linear constraints, in a manner that trades off accuracy versus computational complexity. Symbolic simulation results are presented for a fixed network topology under the assumption of Poisson traffic. The objective is to minimize the average power consumption of a node subject to a minimum average throughput constraint that incorporates soft delay guarantees. The results demonstrate that a MAC protocol that incorporates acknowledgments in a multi-hop setting under dynamic traffic can be generated automatically.

We develop a spatial framework to provide end-to-end delay estimates and guarantees in mobile multihop networks. The novel aspect of this approach is the attribution of network and MAC layer congestion to space, which enables congestion-aware routing and provides delay guarantees over a much longer duration than that achieved by the routing algorithms based on individual nodes. In a mathematically rigorous setting, first, we prove that over the duration during which the node density and the traffic pattern remain stationary, the expected values of local congestion and end-to-end delay roughly remain invariant. Second, we present an accurate method of delay estimation over geographic paths, namely path integration, and derive an upper bound for its estimation error. Further, we develop a congestion-aware routing protocol that utilizes the spatial maps of network congestion to enable delay-optimized routing of real-time applications. Through extensive QualNet simulations, we perform a detailed evaluation of the presented framework and the routing protocol in a realistic setup. The simulation studies demonstrate that the proposed scheme provides substantial improvements in the delivery of real-time applications for a wide range of node densities, velocities, and data traffic.

We formulate the automated MAC protocol generation problem for dynamic topologies, as encountered in wireless ad hoc networks, under multiple neighborhoods and in the presence of acknowledgments. The probability distribution over the set of local topologies encountered in the global network serves as a model for which an optimization program may be formulated that takes the per-node average throughput as its objective function. Since the state space of the model is very large, the main challenge is the generation of the optimization program via automated methods. Symbolic Monte Carlo simulation is used to generate the optimization program, which is subsequently solved via state-of-the-art nonlinear solvers. A quantitative comparison with RTS/CTS provides information on the value of side information on the probability distribution of local topologies, which RTS/CTS does not presume. The results on computational complexity show that the time to generate the program dominates over the time to solve the resulting non-linear program, and that the complete program can be solved within reasonable computational time.

We present a framework for automated Medium Access Control (MAC) protocol generation based on a model that incorporates multiple MAC neighborhoods as well as acknowledgments. Both the advertisement and acknowledgment packets are automatically generated by an optimization program that is built based on symbolic Monte Carlo simulation. The design flow chain produces an optimal MAC protocol with respect to any desired objective function. The simulation results show that RTS/CTS control packets are automatically generated as the topology changes from single neighborhood to multi-neighborhood, and display fast convergence to the optimal solution within very reasonable computational time. This approach has the potential to change the way that MAC protocols are designed from "hand-designed protocols" to protocols designed via design automation chains.

We present a novel framework for spatial routing through difficult terrains with possibly untrustworthy regions in tactical mobile ad hoc networks. We describe a protocol, named "Reliability Map Routing" (RMR), which uses end-to-end Quality of Service metrics that include spatial reliability and trust. The RMR protocol reactively discovers routes over spatial cells whose local reliabilities are distributed throughout the network via a fast dissemination algorithm. By using a spatial approach where reliability and trust are attributed to space rather than to nodes, RMR is able to find reliable routes that persist for longer durations than those discovered by node-centric protocols. Furthermore, RMR is capable of reliable geocasting with low overhead. Via QualNet simulation studies, we compare the performance of the RMR protocol in terms of packet delivery ratio, delay, and overhead, and quantify the effects of node density, velocity, and traffic load on protocol performance. Our results indicate that the RMR protocol performs well in high mobility, high density scenarios because of its controlled routing overhead and spatial approach to routing.

We present an empirical model of the upstream end-to-end Voice-over-IP (VoIP) delay over a mobile, multi-hop wireless network towards a single gateway. First, we define the “VoIP regime” as the regime of network operation in which a fraction 1 - ϵ of the packets are received by the gateway within the maximum tolerable VoIP networking delay y voip , and such that e is less than a maximum tolerable outage probability. We show that within the VoIP regime, the upstream VoIP delay is well-modeled by an exponential distribution whose parameter λ i depends on the number of hops i to the gateway. Second, we show that the coherence time of the VoIP regime is large enough such that the estimation of the {λ i } can be performed by the gateway, and distributed to the nodes such that each node can utilize the {λ i } in estimating the end-to-end delay from its current location for autonomous call admission decisions. Third, we validate the empirical model across scenarios that range from a flat deployment region, to a deployment region with holes, and to the case where there is background User Datagram Protocol (UDP) data traffic. The results are significant for enabling VoIP over high-mobility, multi-hop wireless networks, and in aiding node-based call admission decisions.

We describe a methodology that incorporates the impact of control information transfer into MAC protocol optimization. We apply this methodology to the problem of a synchronous broadcast MAC channel, to generate the optimal protocol when the objective function is the average network throughput per slot. We describe a recursive procedure for the symbolic generation of the optimization program for any choice of the objective function. We demonstrate that this methodology produces the Random Access Channel and protocols with control information as special cases in the regimes where they are optimal. We examine the scaling of optimal throughput and of computational complexity as a function of the number of nodes and the control lifetime.

We show that even though mobile networks are highly unpredictable when viewed at the individual node scale, the end-to-end quality-of-service (QoS) metrics can be stationary when the mobile network is viewed in the aggregate. We define the coherence time as the maximum duration for which the end-to-end QoS metric remains roughly constant, and the spreading period as the minimum duration required to spread QoS information to all the nodes. We show that if the coherence time is greater than the spreading period, the end-to-end QoS metric can be tracked. We focus on the energy consumption as the end-to-end QoS metric, and describe a novel method by which an energy map can be constructed and refined in the joint memory of the mobile nodes. Finally, we show how energy maps can be utilized by an application that aims to minimize a node's total energy consumption over its near-future trajectory.

This paper establishes a framework for the automated generation of networking protocols. The need for the rapid design of networking protocols in application-specific contexts has increased dramatically over the past five years. Each new application requires high performance in its own domain, as well as a rapid design cycle. Further, changes in physical layer technology quickly render previously high-performance protocols poor performance. Currently, there are no tools to automate the design of networking protocols. This paper addresses the challenge of building such tools. It proposes a methodology and a design chain for the automated generation of networking protocols, with the following novel ideas: (1) Formulation of the network protocol problem such that the exchange of protocol control information can be specified as part of the optimization program, (2) Optimal waveform generation, which specifies the optimal exchange of both control information and data, (3) Protocol Extraction: The extraction of the optimal protocol as a minimal description of the optimal waveforms. This methodology has the potential to dramatically change the future of networking protocol design by moving protocol design to a more abstract level, finding high-performance protocols not easily discovered by human intuition, and reducing the currently long protocol design cycles.

Energy planning and optimization constitutes one of the most significant challenges for high-mobility networks. This paper proposes a novel framework to share, retain and refine end- to-end energy metrics in the joint memory of the nodes, over time scales over which this information can be spread to the network and utilized for energy planning decisions. We construct maps of end-to-end energy metrics that enable energy optimization in high-mobility networks. We show how to (1) compute the spatial derivatives of energy potentials in high-mobility networks, (2) construct energy maps on-demand via path integration methods, (3) distribute, share, fuse, and refine energy maps over time by information exchange during encounters, (4) allow the nodes to use energy maps for energy planning and optimization in delay- tolerant, high-mobility networks.

The MIMO single relay channel consists of a transmitter, a relay, and a receiver, all equipped with multiple antennas, located in an environment with distance-dependent path loss as well as scattering. We describe an optimal and a heuristic algorithm to solve the spatial energy allocation problem for the MIMO single relay channel when the scattering clusters on the direct and relay links are non-overlapping. These algorithms can be applied to existing wireless LAN systems that aim to increase their performance via relays. By simulations, we compare the performances of joint and uniform energy allocations across the spatial channels, and the performances of a direct transmission system and a system that utilizes the relay node

We analyze the core capacity region of portable wireless networks under a concave utility model for a heterogeneous collection of asynchronous applications such as email and file transfers. We prove that the core capacity region is non-empty under the concave utility model. We show that there is no relaying among portable users in the core capacity region under the many-to-one traffic model for a single type of traffic. However, heterogeneous traffic induces cooperation when the channel conditions are matched to the type of traffic delivered. We demonstrate that each node can increase its utility by allocating its energy across different portable network configurations. The results motivate the use of energy-limited relays in next-generation wireless LAN systems.

For a wireless network in which every node is bounded in its energy supply, we define a new concept of network capacity called "bits-per-Joule capacity", which is the maximum total number of bits that the network can deliver per Joule of energy deployed into the network. For a fixed network size, a finite number of information bits is delivered for each source-destination pair, under a fixed end-to-end probability of error constraint. We prove that under the one-to-one traffic model in which every node wants to send traffic to a randomly chosen destination node, the bits-per-Joule capacity of a stationary wireless network grows asymptotically as Omega((N/logN)(q-1)/2), where N is the number of nodes randomly deployed onto the surface of a sphere and q is the path loss exponent. Further, the length of the block codes used grows only logarithmically in N, which indicates manageable decoder complexity as the network scales. The fact that the bits-per-Joule capacity grows with the number of nodes contrasts sharply with the scaling laws that have been derived for throughput capacity and implies that large-scale deployments for energy-limited sensor and ad hoc networks may be suitable for delay-tolerant data applications

We model energy-limited, delay-tolerant wireless networks as non-transferable-utility (NTU) cooperative network flow games. We model the utility of each node as a positive linear function of the number of bits that the node sends as a source and the number of bits that it receives as a destination. Based on the core of a cooperative game, we define the "core capacity region" of these networks as the set of utility vectors that cannot be collapsed by any coalition. We show that the core capacity region is non-empty. This implies that even if the nodes have perfect side information on all of the Joules-per-bit link costs in the entire network, there still exists a network solution such that no subset of the nodes can improve the utilities of all of its members over those in this network solution. We present an algorithm to compute the core capacity region. We show that the only core solution under the many-to-one traffic model with a single sink and stationary nodes is the direct transmission strategy; however, when virtual prices are allowed as signals, cooperative solutions emerge within the core capacity region. We show that the core capacity region is non-empty under pricing. The core capacity region serves as a useful concept in the analysis of cooperative, energy-limited, delay-tolerant networks

The spatial single relay channel consists of a transmitter, a relay and a receiver located in an environment with distance-dependent path loss as well as scattering. We describe an optimal and a heuristic algorithm for achieving the maximum data rate from the transmitter to the receiver on non-overlapping spatial channels. These algorithms can be applied to existing wireless LAN systems that aim to increase their performance via relays. We demonstrate the effects of proximity gain and multiplexing gain in determining the optimal relay position. By simulation, we demonstrate the effects of randomly generated scatterers, scatterer position asymmetries, and the scatterer area on the maximum data rate.

We analyze the core capacity region of portable wireless networks under a concave utility model for a heterogeneous collection of asynchronous applications such as email and file transfers. We prove that the core capacity region is non-empty under the concave utility model. We show that there is no relaying among portable users in the core capacity region under the many-to-one traffic model for a single type of traffic. However, heterogeneous traffic induces cooperation when the channel conditions are matched to the type of traffic delivered. We demonstrate that each node can increase its utility by allocating its energy across different portable network configurations. The results motivate the use of energy-limited relays in next-generation wireless LAN systems.

The utility of a node in an energy-limited wireless ad hoc network is defined as a positive linear function of the number of bits that the node sends as a source and the number of bits that it receives as a destination. First, we show that under the one-to-one traffic model in which every node wants to send traffic to a randomly chosen destination node, a utility that grows asymptotically at least as c/sub n/(N/logN)/sup (n-1)/2/ is achievable for every node when the nodes are distributed randomly on the surface of a fixed sphere. In this expression, N denotes the number of nodes, n denotes the transmit power fall-off exponent, and c/sub n/ is a constant that depends on n and is independent of N. Second, we introduce a "dollars-per-Joule pricing" system for wireless ad hoc networks, in which each node can charge any other node a price per Joule of energy that it expends on the other node's traffic. Under this pricing system, we extend the definition of the utility of a node to include the revenue that the node raises and the payments that it makes on the network. We show that the core capacity region of a wireless ad hoc network is non-empty under this dollars-per-Joule pricing system. Further, we show that there exists a sequence of utility vectors in the core capacity region such that the average of the utilities of the nodes grows asymptotically at least as c/sub n/(N/logN)/sup (n-1)/2/ under the one-to-one traffic model.

We develop a framework to evaluate the capacity of wireless ad hoc networks in which the energy supply of the nodes is the primary resource constraint. The measure of capacity in such networks is the "bits-perJoule capacity" that specifies the maximum number of bits that the network can deliver per Joule of energy in the network. First, for the one-to-one traffic model in which each node generates traffic for exactly one other randomly picked node, we demonstrate that the bits-per-Joule capacity grows with the number of nodes. Second, we demonstrate that sparse and locally-computed network topologies can approximate this capacity. Third, for ad hoc-cellular "hybrid" networks in which the node set is partitioned between an ad hoc part and a cellular part, we demonstrate under three representative traffic models, that the ad hoc topology performs far better than the cellular topology.

We define the core capacity region of energy-limited wireless ad hoc networks and show that it is non-empty under the linear utility model. The main motivation for this paper is the question: "Can wireless ad hoc networks grow into a single grand wireless network in the future?" We answer this question by developing a framework in which we incorporate the incentives of users into the definition of network capacity.

In this paper, we evaluate the performance of the minimum energy wireless network, which seeks to provide global connectivity in an ad hoc network while maintaining overall minimum energy for communications. The properties and advantages of employing multi-hop communication over a direct peer-to-peer communication system are illustrated and the performance of such a network is characterized. The blockage percentage for throughput and the overall power consumption are also quantified via simulation.

We describe a distributed position-based network protocol optimized for minimum energy consumption in mobile wireless networks that support peer-to-peer communications. Given any number of randomly deployed nodes over an area, we illustrate that a simple local optimization scheme executed at each node guarantees strong connectivity of the entire network and attains the global minimum energy solution for stationary networks. Due to its localized nature, this protocol proves to be self-reconfiguring and stays close to the minimum energy solution when applied to mobile networks. Simulation results are used to verify the performance of the protocol.

We describe a distributed position-based network protocol optimized for minimum energy consumption in mobile wireless networks that support peer-to-peer communications. Given any number of randomly deployed nodes over an area, we show that a simple local optimization scheme executed at each node guarantees strong connectivity of the entire network and attains the global minimum energy solution for the stationary case. Due to its localized nature, this protocol proves to be self-reconfiguring and stays close to the minimum energy solution when applied to the case of mobile nodes. Our simulations verify its performance.