The all-to-all collective communications primitive is widely used in machine learning (ML) and high performance computing (HPC) workloads, and optimizing its performance is of interest to both ML and HPC communities. All-to-all is a particularly challenging workload that can severely strain the underlying interconnect bandwidth at scale. This paper takes a holistic approach to optimize the performance of all-to-all collective communications on supercomputer-scale direct-connect interconnects. We address several algorithmic and practical challenges in developing efficient and bandwidth-optimal all-to-all schedules for any topology and lowering the schedules to various runtimes and interconnect technologies. We also propose a novel topology that delivers near-optimal all-to-all performance.

The growing demand for edge computing resources, particularly due to increasing popularity of Internet of Things (IoT), and distributed machine/deep learning applications poses a significant challenge. On the one hand, certain edge service providers (ESPs) may not have sufficient resources to satisfy their applications according to the associated service-level agreements. On the other hand, some ESPs may have additional unused resources. In this paper, we propose a resource-sharing framework that allows different ESPs to optimally utilize their resources and improve the satisfaction level of applications subject to constraints such as communication cost for sharing resources across ESPs. Our framework considers that different ESPs have their own objectives for utilizing their resources, thus resulting in a multi-objective optimization problem. We present an${N}$-person Nash Bargaining Solution (NBS) for resource allocation and sharing among ESPs with Pareto optimality guarantee. Furthermore, we propose a distributed, primal-dual algorithm to obtain the NBS by proving that the strong-duality property holds for the resultant resource sharing optimization problem. Using synthetic and real-world data traces, we show numerically that the proposed NBS based framework not only enhances the ability to satisfy applications’ resource demands, but also improves utilities of different ESPs.

Mobile edge computing seeks to provide resources to different delay-sensitive applications. This is a challenging problem as an edge cloud-service provider may not have sufficient resources to satisfy all resource requests. Furthermore, allocating available resources optimally to different applications is also challenging. Resource sharing among different edge cloud-service providers can address the aforementioned limitation as certain service providers may have resources available that can be “rented” by other service providers. However, edge cloud service providers can have different objectives orutilities. Therefore, there is a need for an efficient and effective mechanism to share resources among service providers, while considering the different objectives of various providers. We model resource sharing as a multi-objective optimization problem and present a solution framework based onCooperative Game Theory(CGT). We consider the strategy where each service provider allocates resources to its native applications first and shares the remaining resources with applications from other service providers. We prove that for a monotonic, non-decreasing utility function, the game is canonical and convex. Hence, thecoreis not empty and the grand coalition is stable. We propose two algorithms,Game-theoretic Pareto optimal allocation(GPOA) andPolyandrous-Polygamous Matching based Pareto Optimal Allocation(PPMPOA) that provide allocations from the core. Hence the obtained allocations areParetooptimal and the grand coalition of all the service providers is stable. Experimental results confirm that our proposed resource sharing framework improves utilities of edge cloud-service providers and application request satisfaction.

We address energy-efficient placement of data and analytics components of composite analytics services on a wireless network to minimize execution-time energy consumption (computation and communication) subject to compute, storage and network resource constraints. We introduce an expressive analytics service hypergraph model for representing k-ary composability relationships (k ≥ 2) between various analytics and data components and leverage binary quadratic programming (BQP) to minimize the total energy consumption of a given placement of the analytics hypergraph nodes on the network subject to resource availability constraints. Then, after defining a potential energy functional Φ(·) to model the affinities of analytics components and network resources using analogs of attractive and repulsive forces in physics, we propose a decentralized Metropolis Monte Carlo (MMC) sampling method which seeks to minimize Φ by moving analytics and data on the network. Although Φ is non-convex, using a potential game formulation, we identify conditions under which the algorithm provably converges to a local minimum energy equilibrium placement configuration. Trace-based simulations of the placement of a deep-neural-network analytics service on a realistic wireless network show that for smaller problem instances our MMC algorithm yields placements with total energy within a small factor of BQP and more balanced workload distributions; for larger problems, it yields low-energy configurations while the BQP approach fails.

A basic information theoretic model for summarization is formulated. Here summarization is considered as the process of taking a report of v binary objects, and producing from it a j element subset that captures most of the important features of the original report, with importance being defined via an arbitrary set function endemic to the model. The loss of information is measured by a weight average of variational distances, which we term the semantic loss.Our results include both cases where the probability distribution generating the v-length reports are known and unknown. In the case where the generating distribution is known, our results demonstrate how to construct minimal semantic loss summarizers. For the case where the probability distribution is unknown, we show how to construct summarizers which minimize the semantic loss averaged uniformly over all possible distribution converges to the minimum.

We consider assignment policies that allocate resources to users, where both resources and users are located on a one-dimensional line (0, ∞). First, we consider unidirectional assignment policies that allocate resources only to users located to their left. We propose the Move to Right (MTR) policy, which scans from left to right assigning the nearest available resource located to the right of a user, and contrast it to the Unidirectional Gale-Shapley (UGS) matching policy. While both policies among all unidirectional policies, minimize the expected distance traveled by a request, MTR is fairer. Moreover, we show that when user and resource locations are modeled by statistical point processes, and resources are allowed to satisfy more than one user, the spatial system under unidirectional policies can be mapped into bulk service queueing systems, thus allowing the application of many queueing theory results that yield closed form expressions. As we consider a case where different resources can satisfy different numbers of users, we also generate new results for bulk service queues. We also consider bidirectional policies where there are no directional restrictions on resource allocation and develop an algorithm for computing the optimal assignment which is more efficient than known algorithms in the literature when there are more resources than users. Finally, numerical evaluation of performance of unidirectional and bidirectional allocation schemes yields design guidelines beneficial for resource placement.

Network centrality plays an important role in many applications. Central nodes in social networks can be influential, driving opinions and spreading news or rumors. In hyperlinked environments, such as the Web, where users navigate via clicks, central content receives high traffic, becoming target for advertising campaigns. While there is an extensive amount of work on centrality measures and their efficient computation, controlling nodes' centrality via network updates is a more recent and challenging task. Performing minimal modifications to a network to achieve a desired property falls under the umbrella of network design problems. This paper is focused on improving group (coverage and betweenness) centrality, which is a function of the shortest paths passing through a set of nodes, by adding edges to the network. Several variations of the problem, which are NP-hard as well as APX-hard, are introduced. We present a greedy algorithm, and even faster sampling algorithms, for group centrality maximization with theoretical quality guarantees under realistic constraints. The experimental results show that our sampling algorithms outperform the best baseline solution in terms of centrality by up to 5 times while being 2–3 orders of magnitude faster than our greedy approach.

We study the problem of network topology design within a set of policy-compliant topologies as a game between a designer and an adversary. At any time instant, the designer aims to operate the network in an optimal topology within the set of policy compliant topologies with respect to a desired network property. Simultaneously, the adversary counters the designer trying to force operation in a suboptimal topology. Specifically, if the designer and the attacker choose the same link in the current topology to defend/grow and attack, respectively, then the latter is thwarted. However, if the defender does not correctly guess where the attacker is going to attack, and, hence, acts elsewhere, the topology reverts to the best policy-compliant configuration after a successful attack. We show the existence of various mixed strategy equilibria in this game and systematically study its structural properties. We study the effect of parameters, such as probability of a successful attack, and characterize the steady state behavior of the underlying Markov chain. While the intuitive adversarial strategy here is to attack the most important links, the Nash equilibrium strategy is for the designer to defend the most crucial links and for the adversary to focus attack on the lesser crucial links. We validate these properties through two use cases with example sets of network topologies. Next, we consider a multi-stage framework where the designer is not only interested in the instantaneous network property costs but a discounted sum of costs over many time instances. We establish structural properties of the equilibrium strategies in the multi-stage setting, and also demonstrate that applying algorithms based on the Q-Learning and Rollout methods can result in significant benefits for the designer compared with strategies resulting from a one-shot based game.

We consider the network-wide broadcast problem in multihop wireless networks with reliable multicast at the Medium Access Control (MAC) layer, where the cost of transmitting to downstream nodes at each branch point in the broadcast tree depends on the number$k$of recipients, specifically$1+A k^b$in our model (for some$b \geq 0$,$A \geq 0$). This allows us to capture a wide array of MAC-layer approaches and their costs, simply by varying the value of$b$(relative to$A$), in a problem formulation subsuming the Connected Dominating Set and Spanning Tree problems. We give a systematic analysis of this problem, including positive and negative results. In particular, we show the problem is: approximable by a factor varying from$2H_{\Delta}+2$down to 2 as$b$varies from 0 to 1 (where$\Delta$is the maximum degree of the network graph and$H_\Delta$is the$\Delta$th harmonic number); approximable by a factor varying from 2 to 1 (i.e., optimal) as$b$varies from 1 to$\log _2 (\frac{1}{A}+2)$; and optimally solvable thereafter. Finally, we present numerical results comparing the two algorithms above with other natural heuristics. We find there is an advantage in algorithms taking into consideration the value$b$, even if$b$can only be roughly estimated.

This paper studies wireless networks with multiple real-time flows that have stringent requirements on both per-packet delay and long-term average delivery ratio. Each flow dynamically adjusts its traffic load based on its observation of network status. When the requirements of per-packet delay and delivery ratio are satisfied, each flow obtains some utility based on its traffic load. We aim to design joint rate control and scheduling policies that maximize the total utility in the system. We first show that the problem of maximizing total utility can be formulated as a submodular optimization problem with exponentially many constraints. We then propose two simple distributed policies that require almost no coordination between different entities in the network. The total utilities under these two policies can be made arbitrarily close to the theoretical upper-bound. Extensive simulations also show that they achieve much better performance than state-of-the-art policies.

We study the interactive compression of an arbitrary function of two discrete sources with zero-error. The information on the joint distribution of the sources available at the two sides is asymmetric, in that one user knows the true distribution, whereas the other user observes a different distribution. This paper considers the minimum worst-case zero-error codeword length under such asymmetric prior distributions. We investigate the cases for which reconciling the information mismatch is better or worse than not reconciling it, but instead using an encoding scheme that ensures zero-error with possibly increased communication rate. Our results indicate a reconciliation-communication tradeoff and that there exist cases for which partially reconciling the mismatched information is better than both perfect reconciliation and no reconciliation.

Detecting a small number of outliers from a set of data observations is always challenging. This problem is more difficult in the setting of multiple network samples, where computing the anomalous degree of a network sample is generally not sufficient. In fact, explaining why a given network is exceptional, expressed in the form of subnetwork, is also equally important. We develop a novel algorithm to address these two key problems. We treat each network sample as a potential outlier and identify subnetworks that help discriminate it from nearby samples. The algorithm is developed in the framework of network regression combined with the constraints on both network topology and L1-norm shrinkage to perform subnetwork discovery. Our method thus goes beyond subspace/subgraph discovery. We also show that the developed method converges to a global optimum. Empirical evaluation on various real-world network datasets demonstrates the advantages of our algorithm over various baseline methods.

Modeling information that resides on vertices of large graphs is a key problem in several real-life applications, ranging from social networks to the Internet-of-things. Signal Processing on Graphs and, in particular, graph wavelets can exploit the intrinsic smoothness of these datasets in order to represent them in a compact and accurate manner. However, how to discover wavelet bases that capture the geometry of the data with respect to the signal as well as the graph structure remains an open problem. In this paper, we study the problem of computing graph wavelet bases via sparse cuts in order to produce low-dimensional encodings of data-driven bases. This problem is connected to known hard problems in graph theory (e.g. multiway cuts) and thus requires an efficient heuristic. We formulate the basis discovery task as a relaxation of a vector optimization problem, which leads to an elegant solution as a regularized eigenvalue computation. Moreover, we propose several strategies in order to scale our algorithm to large graphs. Experimental results show that the proposed algorithm can effectively encode both the graph structure and signal, producing compressed and accurate representations for vertex values in a wide range of datasets (e.g. sensor and gene networks) and significantly outperforming the best baseline.

We study the problem of network topology design within a sequence of policy-compliant topologies as a game between a designer and an adversary. At any time instant, the designer aims to operate the network in an optimal topology within this policy compliant sequence with respect to a desired network property. Simultaneously, the adversary counters the designer trying to force operation in a suboptimal topology. We show the existence of various mixed strategy equilibria in this game and systematically study its structural properties. We study the effect of parameters, and characterize the steady state behavior of the underlying Markov chain. While the intuitive adversarial strategy here is to attack links appearing early in the topology sequence, the Nash Equilibrium strategy is for the designer to defend the earlier links and for the adversary to attack the later links. We validate these properties through two use cases with example sets of network topologies.

Real-world networks often consist of multiple layers, be they infrastructure such as airline networks or social such as collaboration networks. A common aspect to these networks is that there are multiple sub-networks that evolve in parallel on the same node set -- these are referred to as multiplex networks. For example, in the case of airline networks, the cities (nodes) have been well-established for several decades if not centuries, but over time new airlines (sub-networks) emerge and each airline creates its own flight linkages between cities. Similarly multiple modalities of communications evolve in parallel between individuals (nodes) such as E-mail, SMS, and Online Social Networks, e.g., Facebook and Twitter. While in some multiplex networks, each layer evolves independently from other layers over time, in other multiple networks, the evolution of a layer is coupled with that of other layers -- a process referred to as co-evolution.

In an ever-increasing variety of contexts, people are working collaboratively to solve problems and accomplish tasks. Yet the characteristics of teams that work effectively are not fully understood. We focus on the problem of identifying particularly effective teams in a large, complex, social network. More specifically, given some task, and taking into account measures of both the effectiveness of an individual and the strength of the pairwise ties between individuals, can we identify the subgroup of people who are most likely to accomplish said task? Our experimental work using the DBLP suggests that within any given subgroup, not all ties are of equal relevance. In fact, we show that focusing on the ties between one individual and the other group members will suffice. However, whereas the problem of finding the "best" subgroup is equivalent to MAX-CLIQUE and is thus hard to approximate, the problem of finding the best star is computationally tractable. We present experimental evidence justifying interest in the star, as opposed to the clique, and discuss algorithmic and complexity concerns.

We study the problem of maximizing the multicast throughput in a dense multi-channel multi-radio (MC-MR) wireless network with multiple multicast sessions. Specifically, we consider a fully connected network topology where all nodes are within transmission range of each other. In spite of its simplicity, this topology is practically important since it is encountered in several real-world settings. Further, a solution to this network can serve as a building block for more general scenarios that are otherwise intractable. For this network, we show that the problem of maximizing the uniform multicast throughput across multiple sessions is NP-hard. However, its special structure allows us to derive useful upper bounds on the achievable uniform multicast throughput. We show that an intuitive class of algorithms that maximally exploit the wireless broadcast feature can result in very poor worst case performance. Using a novel group splitting idea, we then design two polynomial time approximation algorithms that are guaranteed to achieve a constant factor of the throughput bound under arbitrary multicast group memberships. These algorithms are simple to implement and provide interesting tradeoffs between the achievable throughput and the total number of transmissions used.

Timely and efficient message transmission through intermittently and sparsely connected networks is a problem of significant interest to the mobile networking community. Although the long-term statistics describing the time-varying connectivity in such networks can be characterized systematically and can be used for selecting good routes, it may be possible to achieve better performance by intelligently using the actual link states at the time of routing, in conjunction with these statistical dynamics models. In this paper, we investigate a family of minimum expected latency routing methods for such dynamic networks, spanning purely model-based and state-oblivious source routing, state-based source routing, and various flavors of dynamic (or hop-by-hop) routing, with increasing amounts of current link state knowledge around the source. First, we give a heuristic and an approximation scheme for the model-assisted source routing problem, as well as heuristics for the dynamic routing problem. Then we show using extensive simulations on both synthetic and real time-varying connectivity traces that although dynamically sampling link states helps to improve expected routing latency compared to source routing, the marginal improvements decline rapidly for knowledge of current link states beyond 2 hops. To the best of our knowledge, this is the first thorough characterization of the performance of the entire spectrum of model-assisted routing algorithms ranging from little knowledge to complete knowledge of link dynamics.

Time-varying graphs (T-graph) consist of a time-evolving set of graph snapshots (or graphlets). A T-graph property with potential applications in both computer and social network forensics is T-reachability, which identifies the nodes reachable from a source node using the T-graph edges over time period T. In this paper, we consider the problem of estimating the T-reachable set of a source node in two different settings - when a time-evolution of a T-graph is specified by a probabilistic model, and when the actual T-graph snapshots are known and given to us offline (“data aware” setting). Since the value of T could be large in many applications, we propose two simple techniques, namely T-graph sampling and T-graph smashing for significantly reducing the complexity of this computation, while minimizing the estimation error. We show that for the data-aware case, both T-graph sampling and smashing problems are NP-hard, but they are amenable to reasonably good approximations. We also show that for the probabilistic setting where each graphlet in a T-graph is an Erdos-Renyi random graph, sampling yields a loose lower bound for the T-reachable set, while different styles of smashing yield more useful upper and lower bounds. Finally, we show that our algorithms (both data-aware and data-oblivious) can estimate the T-reachable set in real world time-varying networks within reasonable accuracy using less than 0.5% of the number of graphlets.

This paper presents the design and implementation of PUMA, a declarative constraint-solving platform for policy-based routing and channel selection in multiradio wireless mesh networks. In PUMA, users formulate channel selection policies as optimization goals and constraints that are concisely declared using the Colog declarative language. To efficiently execute Colog programs in a distributed setting, PUMA integrates a high-performance constraint solver with a declarative networking engine. We demonstrate the capabilities of PUMA in defining distributed protocols that cross-optimize across channel selection and routing. We have developed a prototype of the PUMA system that we extensively evaluated in simulations and on the ORBIT testbed. Our experimental results demonstrate that PUMA can flexibly and efficiently implement a variety of centralized and distributed channel selection protocols that result in significantly higher throughput compared to single-channel and identical-channel assignment solutions.

The understanding of how humans move is a longstanding challenge in the natural science. An important question is, to what degree is human behavior predictable? The ability to foresee the mobility of humans is crucial from predicting the spread of human to urban planning. Previous research has focused on predicting individual mobility behavior, such as the next location prediction problem. In this paper we study the human mobility behaviors from the perspective of network science. In the human mobility network, there will be a link between two humans if they are physically proximal to each other. We perform both microscopic and macroscopic explorations on the human mobility patterns. From the microscopic perspective, our objective is to answer whether two humans will be in proximity of each other or not. While from the macroscopic perspective, we are interested in whether we can infer the future topology of the human mobility network. In this paper we explore both problems by using link prediction technology, our methodology is demonstrated to have a greater degree of precision in predicting future mobility topology.

We model the problem of channel assignment in mobile networks as one of temporal coloring (T-coloring), that is, coloring a time-varying graph. In order to capture the impact of channel re-assignments due to mobility, we model the cost of coloring as C + αA, where C is the total number of colors used and A is the total number of color changes, and α is a user-selectable parameter reflecting the relative penalty of channel usage and re-assignments.

A recent topic of research in the network science community is combined or composite networks - these are two or more interacting networks that must be characterized jointly rather than individually. For example, a social network and a communication network sharing some nodes (corresponding to users) may be modeled together as a composite network. This may be useful since the aggregate performance of a composite network may often depend on how the individual networks influence each other. Information may travel faster or slower through composite networks depending on how they are coupled. The above vision was captured in the Call for Papers for this special issue in the IEEE Journal On Selected Areas In Communications, and it was published in June 2012. As a result of this solicitation, we received 62 submissions by the deadline of August 15, 2012. Papers were selected after two rigorous rounds of review. The first round of notifications were sent out on December 17, 2012 to the authors whose papers passed the first round of review. Revised versions of the papers were submitted on January 31, 2013. Finally, after a second round of review, the Guest Editorial board decided on March 12, 2013 to accept 16 high quality papers for publication in this competitive special issue. Papers appearing in this special issue belong to five broad themes, which are not necessarily mutually exclusive: (1) fundamental principles in network science, (2) information propagation models in networks, (3) bringing insights from other genres of networks, (4) economic and game theoretic models, and (5) application of network science principles to communications networking problems.

The application of complex network models to communication systems has led to several important results: nonetheless, previous research has often neglected to take into account their temporal properties, which in many real scenarios play a pivotal role. At the same time, network robustness has come extensively under scrutiny. Understanding whether networked systems can undergo structural damage and yet perform efficiently is crucial to both their protection against failures and to the design of new applications. In spite of this, it is still unclear what type of resilience we may expect in a network which continuously changes over time. In this work, we present the first attempt to define the concept of temporal network robustness: we describe a measure of network robustness for time-varying networks and we show how it performs on different classes of random models by means of analytical and numerical evaluation. Finally, we report a case study on a real-world scenario, an opportunistic vehicular system of about 500 taxicabs, highlighting the importance of time in the evaluation of robustness. Particularly, we show how static approximation can wrongly indicate high robustness of fragile networks when adopted in mobile time-varying networks, while a temporal approach captures more accurately the system performance.

This paper presents Cologne, a declarative optimization platform that enables constraint optimization problems (COPs) to be declaratively specified and incrementally executed in distributed systems. Cologne integrates a declarative networking engine with an off-the-shelf constraint solver. We have developed the Colog language that combines distributed Datalog used in declarative networking with language constructs for specifying goals and constraints used in COPs. Cologne uses novel query processing strategies for processing Colog programs, by combining the use of bottom-up distributed Datalog evaluation with top-down goal-oriented constraint solving. Using case studies based on cloud and wireless network optimizations, we demonstrate that Cologne (1) can flexibly support a wide range of policy-based optimizations in distributed systems, (2) results in orders of magnitude less code compared to imperative implementations, and (3) is highly efficient with low overhead and fast convergence times.

This paper presents DAWN, a declarative platform that creates highly adaptive policy-based mobile ad hoc network (MANET) protocols. DAWN leverages declarative networking techniques to achieve extensible routing and forwarding using declarative languages. We make the following contributions. First, we demonstrate that traditional MANET protocols can be expressed in a concise fashion as declarative networks and policy-driven adaptation can be specified in the same language to dictate the dynamic selection of different protocols based on various network and traffic conditions. Second, we propose interprotocol forwarding techniques that ensure packets are able to seamlessly traverse across clusters of nodes running different protocols selected based on their respective policies. Third, we have developed a full-fledged implementation of DAWN using the RapidNet declarative networking system. We experimentally validate a variety of policy-based adaptive MANETs in various dynamic settings using a combination of ns-3 simulations and deployment on the ORBIT testbed. Our experimental results demonstrate that hybrid protocols developed using DAWN outperform traditional MANET routing protocols and are able to flexibly and dynamically adapt their routing mechanisms to achieve a good tradeoff between bandwidth utilization and route quality. We further demonstrate DAWN's capabilities to achieve interprotocol forwarding across different protocols.

The application of complex network theory to communication systems has led to several important results. Nonetheless, previous research has often neglected to take into account their temporal properties, which in many real scenarios play a pivotal role. Mainly because of mobility, transmission delays or protocol design, a communication network should not be considered only as a static entity. At the same time, network robustness has come extensively under scrutiny. Understanding whether networked systems can undergo structural damage and yet perform efficiently is crucial to both their protection against failures and to the design of new applications. In spite of this, it is still unclear what type of resilience we may expect in a network that continuously changes over time. In this work we present the first attempt to define the concept of temporal network robustness: we describe a measure of network robustness for time-varying networks and we show how it performs on different classes of random models by means of analytical and numerical evaluation. Particularly, we show how static approximation can wrongly indicate high robustness of fragile networks when adopted in mobile time-varying networks, while a temporal approach captures more accurately the system performance.

We consider the broadcasting problem in multi-radio multi-channel ad hoc networks. The objective is to minimize the total broadcast cost, where the cost can be of any form that is summable over all the transmissions (e.g., the transmission and reception energy, the price for accessing a specific channel). Our technical approach is based on a simplicial complex model that allows us to capture the broadcast nature of the wireless medium and the heterogeneity across radios and channels. Specifically, we show that broadcasting in multi-radio multi-channel ad hoc networks can be formulated as a minimum spanning problem in simplicial complexes. We establish the NP-completeness of the minimum spanning problem and propose two approximation algorithms with order-optimal performance guarantee. These two algorithms offer tradeoffs between performance and time complexity. In a broader context, this work appears to be the first that studies the minimum spanning problem in simplicial complexes and weighted minimum connected set cover problem.

In a composite network, a piece of information traveling through links in a social network may have to travel over multiple links in an associated communication network. In this paper, we propose a model of composite networks that consists of two networks and an embedding between them, and several composite metrics that characterize information flow under a particular type of embedding. We present analytic results for the scaling behavior of "constrained composite stretch" of a path, "constrained composite diameter" of a graph, and "constrained composite broadcast time" of a tree, under random uniform embeddings onto various communication network structures. We validate our analytical results on composite stretch using two data sets consisting of a friendship social network geographically spread across Western Europe and a historical deployment of a military chain of command. We also present a randomized model of field deployment consistent with real-world data, and use simulations over this model to explore the distribution of constrained composite broadcast time. Finally, we show that our analytical bounds for composite broadcast time agree well with the simulation results.

Duty cycling or periodic sleep scheduling of RF transceivers of nodes in a wireless ad hoc or sensor network can significantly reduce energy consumption. This paper sheds light on the fundamental limits of the end-to-end data delivery latency and the per-flow throughput in a wireless network with multiple interfering flows, in the presence of "coordinated" duty cycling. We propose green wave sleep scheduling (GWSS) - inspired by synchronized traffic lights - for scheduling sleep-wake slots and routing data in a duty cycling wireless network, whose performance can approach the aforementioned limits. Particularly, we derive a general latency lower bound and show that GWSS is latency optimal on various structured topologies, such as the line, grid and the tree, at low traffic load. For an arbitrary network, finding a solution to the delay-efficient sleep scheduling problem is NP-hard. But for the 2D grid topology, we show that a non-interfering construction of GWSS is optimal in the sense of scaling laws of latency and capacity. Finally, using results from percolation theory, we extend GWSS to random wireless networks, where nodes are placed in a square area according to the Poisson point process. Aided by strong numerical evidence for a new conjecture on percolation on a semi-directed lattice that we propose, we demonstrate the latency optimality of GWSS on a random extended network, i.e., for an area-n random network with unit-density-Poisson distributed nodes, and a node-active (duty-cycling) rate p, GWSS can achieve a per-flow throughput scaling of T(n, p) = Ω(p/√n) bits/sec and latency D(n, p) scaling of O(√n) + O(1/p) hops/packet/flow.

Stateless opportunistic forwarding is a simple fault-tolerant distributed scheme for packet delivery, data gathering, and information querying in intermittently connected networks by which packets are forwarded to the next available neighbors in a “random walk” fashion until they reach their intended destinations or expire. It has been employed in diverse situations, for instance, when: 1) the global network topology is not known or is highly dynamic; 2) the availability of the next-hop neighbors is not easily controllable; or 3) the relaying nodes are computationally constrained. Data delivery in sensor networks, ad hoc networks, and delay-tolerant networks are well-known applications besides searching in peer-to-peer networks. A major challenge for stateless opportunistic forwarding is the difficulty to predict the end-to-end latency. To facilitate latency evaluation, we study a simplified model of stateless opportunistic forwarding, namely a “weighted random walk” in a finite graph. This paper makes several contributions toward the analysis of this model. 1) By spectral graph theory we derive a general formula to efficiently compute the exact hitting and commute times of random walks with heterogeneous transition times at relay nodes. Such transition times can model the heterogeneous delivery times and availability periods of the next-hop neighbors. 2) We study a common class of distance-regular networks with a varying number of geographical neighbors and obtain exact and approximation formulas for the hitting time in such networks. 3) Based on these results, we study other sophisticated settings, such as random geographical locations, topology-aware forwarding, and multicopy random-walk forwarding. Our results provide the basic analytical tools for managing and controlling the performance of stateless opportunistic forwarding in finite networks.

While scheduling the nodes in a wireless network to sleep periodically can save energy, it also incurs higher latency and lower throughput. We consider the problem of designing optimal sleep schedules in wireless networks, and show that finding sleep schedules that can minimize the latency over a given subset of source-destination pairs is NP-hard. We also derive a latency lower bound given by d + O(1/p) for any sleep schedule with a required active rate (i.e., the fraction of active slots of each node) p, and the shortest path length d. We offer a novel solution to optimal sleep scheduling using green-wave sleep scheduling (GWSS), inspired by coordinated traffic lights, which is shown to meet our latency lower bound (hence is latency-optimal) for topologies such as the line, grid, ring, torus and tree networks, under light traffic. For high traffic loads, we propose non-interfering GWSS, which can achieve the maximum throughput scaling law given by T(n,p) = ¿(p/¿n) bits/sec on a grid network of size n, with a latency scaling law D(n,p) = O(¿n) + O(1/p). Finally, we extend GWSS to a random network with n Poisson-distributed nodes, for which we show an achievable throughput scaling law of T(n,p) = ¿(p/¿(n log n)) bits/sec and a corresponding latency scaling law D(n,p) = O(¿(n/log n)) + O(1/p); hence meeting the well-known Gupta-Kumar achievable throughput rate ¿(1/¿(n log n)) when p ¿ 1.

Mobility models have traditionally been tailored to specific application domains such as human, military, or ad hoc transportation scenarios. This tailored approach often renders a mobility model useless when the application domain changes, and leads to wrong conclusions about the performance of protocols and applications running atop of different domains. In this work we propose and implement a mobility modeling framework (UMMF) based on the observation that the mobility characteristics of most mobility-based applications can be captured in terms of a few fundamental factors: (1) Targets; (2) Obstacles; (3) Dynamic events; (4) Navigation; (5) Steering behaviors; and (6) Dynamic behaviors. We demonstrate the mapping of application-domain-specifics to UMMF elements, showing the power and flexibility of our approach.

This paper presents the design and implementation of declarative policy-based adaptive MANET routing protocols. Our work builds upon declarative networking, a recent innovation for building extensible network architectures using declarative languages. We make the following contributions. First, we demonstrate that traditional MANET protocols can be expressed in a compact fashion as declarative networks. We validate these declarative protocols via an experimental study on the ORBIT wireless testbed and a cluster-based emulation environment. Second, we demonstrate that policy-driven adaptation can be specified in a generic set of declarative rule-based policies that dictate the dynamic selection of different protocols based on network conditions. Third, we conduct extensive evaluation results of declarative policy-based adaptation of MANET routing on the ORBIT wireless testbed and the cluster-based emulation environment. Our experimental results show that the specified policies enable MANETs to dynamically hybridize a variety of routing protocols to achieve a good tradeoff in bandwidth utilization and route quality.

Stateless opportunistic forwarding is a simple fault- tolerant distributed approach for data delivery and information querying in wireless ad hoc networks, where packets are forwarded to the next available neighbors in a "random walk" fashion, until they reach the destinations or expire. This approach is robust against ad hoc topology changes and is amenable to computation/bandwidth/energy-constrained devices; however, it is generally difficult to predict the end-to-end latency suffered by such a random walk in a given network. In this paper, we make several contributions on this topic. First, by using spectral graph theory we derive a general formula for computing the exact hitting and commute times of weighted random walks on a finite graph with heterogeneous sojourn times at relaying nodes. Such sojourn times can model heterogeneous duty cycling rates in sensor networks, or heterogeneous delivery times in delay tolerant networks. Second, we study a common class of distance-regular networks with varying numbers of geographical neighbors, and obtain simple estimate-formulas of hitting times by numerical analysis. Third, we study the more sophisticated settings of random geographical locations and distance-dependent sojourn times through simulations. Finally, we discuss the implications of this on the optimization of latency-overhead trade-off.

Existing slotted channel access schemes in wireless networks assume that slot boundaries at all nodes are synchronized. In practice, relative clock drifts among nodes cause slot misalignment over time and can result in catastrophic data loss in such systems. We propose a simple network-wide slot synchronization scheme (Slot-Sync and Slot-Resync) suitable for duty-cycling wireless sensor networks. The proposed scheme attains high accuracy by circumventing dominant sources of error inherent in traditional time synchronization protocols. At moderate duty-cycle frequencies, the proposed scheme also has the unique advantage of eliminating re-synchronization (re-sync) overhead completely, thereby achieving slot re-sync essentially for free. We provide an energy efficient slot guard time and re- sync interval design for the proposed scheme and analyze several spanning tree structures for slot-alignment message propagation. In addition, we derive upper bounds on the synchronization (sync) error for a family of trees. Through simulations, we compare the spanning trees we propose to those used for time sync in literature and show up to 80% reduction in sync error and up to 70% reduction in energy needed for slot-alignment message propagation by choosing appropriate tree structures.

Since battery energy is often a premium resource in wireless sensor networks, many systems attempt to conserve battery energy by turning the transceivers off in a process referred to as duty-cycling. In [4-5], the authors proposed two different random duty-cycling protocols utilizing pseudo random number generators to efficiently exchange ON/OFF schedules among neighboring nodes for transmission coordination. We propose two simple transmission schemes that are superior to the performance of the schemes in [4-5] in terms of energy consumption and a macro network metric termed delivery capacity. The delivery capacity measures the maximum long term average unicast traffic delivery capability of a network. Through analyses and simulations, we present a detailed energy and delivery capacity tradeoff study of the transmission schemes in [4-5] and the schemes we propose. We show that one of our proposed schemes is near-optimal in energy consumption for low traffic loads. In addition, we provide insights on the appropriate ON/OFF duty-cycling parameters that can achieve the optimal energy and delivery capacity tradeoff curve.

Abstract – Data routing in large wireless sensor networks is challenged by requirements for scalability, robustness, and energy use, as well as the possibility that globally unique identifiers for each node are nonexistent. Moreover, if the sensor network serves multiple functions or users, its routing function must be energy efficient when responding to information requests (inquiries) that can arrive at high rates, require diverse data types, and target subsets of the sensor network. We propose virtual “containment ” hierarchies based on relevant attributes as a more efficient mechanism to disseminate inquiries rather than the use of flooding schemes. We show algorithms that support clustering sensors according to these hierarchies and implement clusterhead failure recovery and load balancing among cluster members. We show that our framework has significant bandwidth gains over a flooding-based scheme under the scenario considered using analytical techniques.

Energy efficiency is an important design criterion for the development of sensor networking protocols involving data dissemination and gathering. In-network processing of sensor data, aggregation, transmission power control in radios, and periodic cycling of node wake-up schedules are some techniques that have been roposed in the sensor networking literature for achieving energy efficiency. Owing to the broadcast nature of the wireless channel many nodes in the vicinity of a sender node may overhear its packet transmissions even if they are not the intended reci ients of these transmissions. Reception of these transmissions can result in unnecessary expenditure of battery energy of the recipients. In this paper,we investigate the impact of overhearing transmissions on total energy costs during data gathering and dissemination and attempt to minimize them systematically.We model the minimum energy data gathering problem as a directed minimum energy spanning tree problem where the energy cost of each edge in the wireless connectivity graph is augmented by the overhearing cost of the corresponding transmission. We observe that in dense sensor networks, overhearing costs constitute a significant fraction of the total energy cost and that computing the minimum spanning tree on the augmented cost metric results in energy savings, especially in networks with non-uniform s atial node distribution. We also study the impact of this new metric on the well known energy-efficient dissemination (also called broadcasting) algorithms for multihop wireless networks.We show via simulation that through this augmented cost metric, gains in energy efficiency of 10%or more are ossible without additional hardware and minimal additional complexity.

We propose a task graph based framework for modeling and execution of distributed applications in mobile ad hoc networks. Our framework represents a distributed application by a graph composed of nodes and edges in which the nodes logically represent application sub-tasks that need to be completed and the edges represent associations, with certain attributes, between nodes. During application run-time, suitable devices that can complete the sub-tasks and can satisfy the attributes of the associations between them are selected on-the-fly to execute the application. New devices are selected to continue application execution if old devices become unavailable due to mobility. Thus, we decouple the application from a specific set of devices and allow its execution if there is at least one suitable device in the network for carrying out each of the required sub-tasks. We propose an application execution protocol to realize this vision and show simulation results which indicate that our approach is practical for environments with low user/device mobility.

A novel distributed approach for executing distributed tasks on mobile ad hoc networks (MANETs) is presented. A distributed application is represented as a complex task comprised of simpler sub-tasks that need to be performed on different categories of computing devices with specialized roles. The dependencies induced by logical patterns of data flow between these specialized devices which are responsible for performing the aforementioned sub-tasks, yield a task graph representation for a given application. We present a simple and efficient distributed algorithm for dynamic discovery and selection of suitable devices in a MANET from among a number of them providing the same functionality. We refer to this process as instantiation which is carried out with respect to the proposed task graph representation of the application. We also present a distributed algorithm for detecting disruptions in application execution that can occur due to device mobility. The algorithm then recovers quickly from the situation by re-instantiating affected parts of the task graph, if possible. Finally, we propose metrics for evaluating the performance of these algorithms and report simulation results for a variety of application scenarios.

Video-on-demand (VoD) has been an active area of research for the past few years in the multimedia research community. However, there have not been many significant commercial deployments of VoD owing to the inadequacy of per user bandwidth and the lack of a good business model.1 Significant research efforts have been directed towards reduction of network bandwidth requirements, improvement of server utilization, and minimization of start-up latency. In this paper, we investigate another aspect of VoD systems which has been largely neglected by the research community, namely, pricing models for VoD systems. We believe that the price charged to a user for an on-demand video stream should influence the rate of user arrivals into the VoD system and in turn should depend upon quality-of-service (QoS) factors such as initial start-up latency. We briefly describe some simple pricing models and analyze the tradeoffs involved in such scenarios from a profit maximization point of view. We further explore secondary content insertion (ad-insertion) which was proposed elsewhere [1] not only as a technique for reducing the resource requirements at the server and the network, but also as a means of subsidizing VoD content to the end user. We treat the rate of ad insertion as another QoS factor and demonstrate how it can influence the price of movie delivery.

We present and evaluate an optimal scheduling algorithm for inserting secondary content for improving resource utilization in VoD systems. The algorithm runs in polynomial time, and is optimal with respect to the total bandwidth usage over the merging interval. We present constraints on content insertion which make the overall QoS of the delivered stream acceptable, and show how our algorithm can satisfy these constraints. We discuss dynamic scenarios with user arrivals and interactions, and show by simulations that content insertion reduces the channel bandwidth requirement to almost half.