In hybrid datacenter networks, wired connections are augmented with wireless links to facilitate data transfers between racks. The usage of mmWave/FSO wireless links enables dynamic bandwidth/capacity allocation with extremely small reconfiguration delay. Also, on-demand workload distribution, where the workload of a job is divided into multiple tasks that can be distributed/routed to different racks to be processed in parallel, allows better utilization of computational resources in data centers. In prior work, the dynamic wireless capacity augmentation and workload distribution decisions were mostly made independently and in a heuristic manner for serving distributed and parallel computing jobs. In this paper, we propose a novel analytical framework and algorithms to jointly optimize both the wireless capacity augmentation and the workload distribution, to minimize the job completion time. We consider workload that is not amenable to pipelining, fully amenable to pipelining, and partially amenable to pipelining. With extensive simulation studies, we show that the gain (in terms of the reduction in the job completion time) can be very substantial when allowing such joint optimization.

In large-scale cloud computing systems, a task is often divided into multiple subtasks which can be executed in parallel in different machines. As a result, the task completion time is constrained by the completion time of the slowest subtask. To reduce the task completion time, the strategy of replicating the straggling subtasks has been employed in cloud computing frameworks such as MapReduce and Hadoop. Analyzing mathematically the performance of such replication strategies has recently received great attention. However, most of the analytical work focuses on the case where the completion times of the subtasks are identically distributed. This assumption may not hold in practice due to the modularization and encapsulation of the computation of a task, resulting in different service requirements for different subtasks. In this paper, we consider the case where the completion times of the subtasks of a task are drawn from heterogeneous/empirical distributions. Furthermore, we consider the case where jobs consist of hierarchical tasks that are required to be executed in a specific order described by a task precedence graph. We propose a novel framework to investigate how to allocate replication resources among the subtasks such that the overall task completion time is minimized. Specifically, we devise a Lagrange multiplier-based method and a water-filling-like algorithm for integer programs. We show via analysis and simulations the optimality and efficiency of our proposed algorithms, and explore the tradeoff between cost and latency from introducing replications in a task graph.

The demand for higher and higher wireless data rates is driven by the popularity of mobile video content delivery through wireless devices such as tablets and smartphones. To achieve unprecedented mobile content delivery speeds while reducing backhaul cost and delay, in this paper we propose a new system architecture that combines two recent ideas, distributed caching of content in small cells (FemtoCaching), and, cooperative transmissions from nearby base stations (Coordinated Multi-Point). A key characteristic of the proposed architecture is the interdependence between the caching strategy and the physical layer coordination. Specifically, the caching strategy may cache different content in nearby base stations (BSs) to maximize the cache hit ratio, or cache the same content in multiple nearby BSs such that the corresponding BSs can transmit concurrently, e.g., to multiple users using zero-forcing beamforming, and achieve multiplexing gains. Such interdependency allows a joint cross-layer optimization. Given the popularity distribution of the content, the available cache size, and the network topology, we devise near-optimal strategies of caching such that the system throughput is maximized or the system delay is minimized. Under realistic scenarios and assumptions, our analytical and simulation results show that our system yields significantly faster content delivery, which can be one order of magnitude faster than that of legacy systems.

Modern wireless devices such as smartphones are pushing the demand for higher wireless data rates. The ensuing increase in wireless traffic demand can be met by a denser deployment of access points, coupled with a coordinated deployment of advanced physical layer techniques to reduce inter-cell interference. Unfortunately, advanced physical layer techniques, e.g., multi-user (MU) MIMO found in 802.11ac and LTE-advanced, are not designed to operate efficiently in a coordinated fashion across multiple densely deployed transmitters. In this paper, we introduce a new coordination architecture, which can achieve high performance gains without the high overhead and deployment cost that usually comes with coordination, thus making the vision of high capacity wireless access via densely deployed transmitters practical. The basic idea is to loosely coordinate nearby transmitters using slow varying channel statistics, while keeping all the functionality which depends on fast varying channel state information and has tight time deadlines locally. We achieve this via a smart combination of analog and digital beamforming using inexpensive front ends, a provably efficient algorithm to select compatible users and analog beams across all transmitters, and backward compatible protocol extensions. Our performance results, which include analysis, simulations, and experiments with software defined radios and directional antennas, show that our approach can achieve the $10\times $ gains of the theoretically optimal coordinated MU-MIMO approach, without the need to either tightly coordinate the clocks of the remote transmitters or meet tight delay constraints.

Cellular base band units of the future are expected to reside in a cloud data center which provides computation resources, content storage and caching, and a natural place to perform multi-user precoding, thus addressing both cost and performance concerns of cellular systems. Multi-user precoding relies on efficient user grouping schemes to maximize multiplexing gains. However, traditional user grouping schemes are unaware of data center constraints, and may induce a large number of data transfers across racks when fetching requested data to a certain rack for precoding. When congestion occurs in the data center network, the delay of data transfers across racks may exceed the channel coherence time. This would kill multi-user MIMO transmissions as channel state information becomes outdated. In this paper, we design a novel data-locality-aware user grouping schemes which preferentially group users whose requested data are located under the same rack. We also design user grouping algorithms which adapt to the congestion level in the cloud data center. Specifically, a regularized spectral efficiency maximization problem is proposed where the number of data transfers across racks is introduced as a regularization term. By adjusting the weight of the regularization term according to the congestion level, we gradually suppress data transfers across racks in forming user groups when congestion occurs. We reduce the above problem to a soft-capacitated facility location problem, and we devise a 2-approximation user grouping algorithm. At last, we conduct simulations which show that our proposed algorithm performs close to the optimal in practical scenarios, and study the tradeoff between higher spectral efficiency and lower data transfer cost.

The constantly growing wireless bandwidth demand is pushing wireless networks to multi-tier architectures consisting of a macrocell tier and a number of dense small cell deployment tiers. In such a multi-tier multi-cell environment, the classic problem of associating users to base stations becomes both more challenging and more critical to the overall network performance. Most previous analytical work is focused on designing static user-cell association algorithms, which, to achieve optimality, are periodically applied whenever there are new user arrivals, thus potentially inducing a large number of re-associations for previously arrived users. On the other hand, practical online algorithms that do not allow any such user re-association are often based on heuristics and may not have any performance guarantees. In this paper, we propose online algorithms for the multi-tier multi-cell user association problem that have provable performance guarantees, which improve previously known bounds by a sizable amount. The proposed algorithms are motivated by online combinatorial auctions, while capturing and leveraging the relative sparsity of choices in wireless networks as compared with auction setups. Our champion algorithm is a 1/2-a-1approximation algorithm, where a is the maximum number of feasible associations for a user and is, in general, small due to path loss. Our analysis considers the state-of-the-art wireless technologies, such as massive and multiuser MIMO, and practical aspects of the system such as the fact that highly mobile users have a preference to connect to larger cell tiers to keep the signaling overhead low. In addition to establishing formal performance bounds, we also conduct simulations under realistic assumptions, which establish the superiority of the proposed algorithm over existing approaches under real-world scenarios.

Technologies for millimeter wave (mmWave) communication are at the forefront of investigations in both industry and academia, as the mmWave band offers the promise of orders of magnitude additional available bandwidths to what has already been allocated to cellular networks. The much larger number of antennas that can be supported in a small footprint at mmWave bands can be leveraged to harvest massive-MIMO type beamforming and spatial multiplexing gains. Similar to Long-Term Evolution (LTE) systems, two prerequisites for harvesting these benefits are detecting users and acquiring user channel state information (CSI) in the training phase. However, due to the fact that mmWave channels encounter much harsher propagation and decorrelate much faster, the tasks of user detection and CSI acquisition are both imperative and much more challenging than in LTE bands. In this paper, we investigate the problem of fast user detection and CSI acquisition in the downlink of small cell mmWave networks. We assume Time Division Duplex (TDD) operation and channel-reciprocity based CSI acquisition. To achieve densification benefits we propose pilot designs and channel estimators that leverage a combination of aggressive pilot reuse with fast user detection at the base station and compressed sensing channel estimation. As our simulations show, the number of users that can be simultaneously served by the entire mmWave-band network with the proposed schemes increases substantially with respect to traditional compressed sensing based approaches with conventional pilot reuse.

The ever growing wireless bandwidth demand is pushing WiFi and cellular networks to dense multi-cell deployments, as well as to multi-tier architectures consisting of macrocells and small cells. In such a multi-tier multi-cell environment, the classic problem of associating users to base stations becomes both more challenging and more critical to the overall network performance. Most previous analytical work is focused on offline/static user-cell association, where the users' arrivals and their rates are assumed to be known in advance and thus has little practical relevance. On the other hand, practical online algorithms based on heuristics are often suboptimal and may not provide any performance guarantees. In this paper, we propose an online algorithm for the multi-tier multi-cell user association problem that has a provable performance guarantee which improves previously known bounds by a sizable amount. The proposed algorithm is motivated by online combinatorial auctions, while capturing and leveraging the relative sparsity of choices in wireless networks as compared to auction setups. Specifically, it is a 1/2−a−1 approximation algorithm, where a is the maximum number of feasible associations for a user and is, in general, small due to path loss. In addition to establishing formal performance bounds, we also conduct simulations under realistic assumptions which establish the superiority of the proposed algorithm over existing approaches under real-world scenarios.

Modern wireless devices such as tablets and smartphones are pushing the demand for higher and higher wireless data rates. The vast majority of this demand comes from media content. In this paper we propose to combine two recent ideas, distributed caching of content in small cells, and, cooperative transmissions from nearby base stations/BSs (generally known as coordinated multi-point), to achieve unprecedented content delivery speeds while reducing backhaul cost and delay. A key characteristic of our architecture is the interdependence between the caching strategy and the PHY/MAC layer coordination. Specifically, the caching strategy may cache different content in nearby BSs to maximize the hit ratio, or cache the same content in multiple nearby BSs such that the corresponding BSs can transmit concurrently, e.g. to multiple users using zero force beamforming, and achieve multiplexing gains. With this in mind, given the popularity distribution of the content, the available cache size, and the network topology, we devise optimal strategies of caching such that the throughput of the system is maximized. Our analytical and simulation results show that our system yields significantly faster content delivery, which, under realistic scenarios and assumptions can be one order of magnitude faster than that of legacy systems.

The spatial aspect of distributed dynamic spectrum access (DSA) in a wireless ad hoc network is usually overlooked in the literature. By using tools from stochastic geometry, we analytically characterized the spatial distribution of users and subsequent interference in the network. Provided with multiple channels, users in the network interact in the sense that the channel selections they make affects co-channel interference they impose on each other, forming a new and general scope of DSA. The study of strategic interactions among users sharing the spectrum are facilitated by a game-theoretic formulation, where the strategy at the Nash equilibrium and the corresponding performance are analytically obtained. Based on the model, we consider further the case when users have multi-channel sensing capability. Access strategies incorporating channel availability information are developed and the corresponding performance is analyzed. Finally, we show that channel (frequency) diversity can be exploited that enables effective design when users are capable of performing concurrent communications over multiple channels. The results of our work provides insights to and facilitates the design of medium access strategies for DSA.

This paper analyzes the delay of multi-hop transmissions in cognitive radio ad hoc networks (CRNs) to investigate the performance of routings in CRN. Compared to routing in ad hoc networks, additional medium access delay is introduced in CRN since opportunistic transmissions of secondary users (SUs) should not violate the interference constraints at primary receivers. Moreover, additional retransmission delay occurs in CRN because received signals at SUs are interfered with by both primary transmitters and concurrent secondary transmitters. We propose an analytically tractable model to investigate routings in CRN considering medium access delay, retransmission delay and the hop count of the end-to-end route. Through optimizing the number of hops of the end-to-end route, a lower bound on the end-to-end packet transmission delay is developed, which facilitates delay QoS provisioning and rate-delay trade-off in CRN. Consequently, this research serves as the valid framework for baseline performance analysis and offers novel avenues to routing design in CRN.

Local (single-hop) broadcasting is widely employed in distributed protocols (e.g., neighbor discovery, local information exchange in distributed network optimization and gossip-based algorithms) in wireless ad hoc networks. The performance of local broadcasting is characterized by the mean number of neighbors and the probability distribution of the number of neighbors of a broadcasting node. In this paper, we study the performance of local broadcasting in heterogeneous wireless ad hoc networks in which inter-system interference dominates signal reception quality, considering general fading distributions of both the desired signal and the interfering signals. In addition, we investigate the impacts of different error control techniques (e.g., simple retransmission, Chase combining, and incremental redundancy) on the performance of local broadcasting. The increase in the mean number of neighbors of a broadcasting node with respect to the number of retransmissions of a message is clarified, facilitating QoS provisioning in reliable local broadcasting in interference-limited heterogeneous wireless ad hoc networks.

Cognitive radio (CR) enables dynamic spectrum access to provide universal connectivity across different types of radio access technologies, realizing seamless content delivery in the next generation wireless networking. As the most general cognitive radio network scenario, multiple ad hoc and infrastructure networks sharing the same spectrum are allowed to cooperate with each other by using other networks' nodes as relays to carry and forward traffic, forming an interconnected heterogeneous network. Previous research is limited to the connectivity analysis of hybrid wireless networks with path loss channel model. In this paper, we analyze the connectivity of the interconnected heterogeneous network consisting of multiple ad hoc networks and multiple infrastructure networks from a percolation-based perspective, considering both noise-limited and interference-limited environment with general fading channel model. The benefit of network-based cooperation is quantified in terms of the percolation threshold, facilitating proper network control and deployment. For example, using such quantification, we can specify the amount of infrastructure deployment or reliable connections to control the connectivity of existing ad hoc networks, and vice versa. Finally, we apply the analysis to the coexistence of primary and secondary networks, as the common CR scenario.

In cognitive radio networks, the signal reception quality of a secondary user degrades due to the interference from multiple heterogeneous primary networks, and also the transmission activity of a secondary user is constrained by its interference to the primary networks. It is difficult to ensure the connectivity of the secondary network. However, since there may exist multiple heterogeneous secondary networks with different radio access technologies, such secondary networks may be treated as one secondary network via proper cooperation, to improve connectivity. In this paper, we investigate the connectivity of such a cooperative secondary network from a percolation-based perspective, in which each secondary network's user may have other secondary networks' users acting as relays. The connectivity of this cooperative secondary network is characterized in terms of percolation threshold, from which the benefit of cooperation is justified. For example, while a noncooperative secondary network does not percolate, percolation may occur in the cooperative secondary network; or when a noncooperative secondary network percolates, less power would be required to sustain the same level of connectivity in the cooperative secondary network.

The scope of this paper is two-fold. First, we extend the concept of transmission capacity of a wireless network to the broadcast scenario and consequently define the broadcast transmission capacity of a wireless network as the product of the spatial density of broadcasting nodes, the average number of neighbors of a broadcasting node, and the information rate, considering both intra-system interference and inter-system interference from multiple coexisting heterogeneous wireless networks. The broadcast transmission capacity is obtained under general fading of desired signal and interfering signals. Second, we characterize the impact of eavesdroppers on the broadcast transmission capacity of a wireless network with secrecy outage constraint. In this case, the information rate is replaced by the secrecy rate. Our results illustrate performance limits of a network coexisting with other heterogeneous networks (and/or eavesdroppers) and trade-offs between different system parameters, and thus suggest network design (e.g., optimal medium access probability) of protocols that require local broadcasting, such as local information exchange in distributed network optimization and gossip-based algorithms.

In cognitive radio networking, the signal reception quality of a secondary user degrades due to the interference from multiple heterogeneous primary networks, and also the transmission activity of a secondary user is constrained by its interference to the primary networks. It is difficult to ensure the 1-connectivity of the secondary network, so we here characterize the connectivity of the secondary network from a percolation-based perspective. On the other hand, since there may exist multiple heterogeneous secondary networks with different radio access technologies, such secondary networks may be treated as one secondary network via proper cooperation. In this paper, we investigate the connectivity of such a cooperative secondary network, under which each secondary network's user may have other secondary networks' users acting as relays. The connectivity of this cooperative secondary network is characterized in terms of percolation threshold, from which the benefits of cooperation are justified. For example, while a noncooperative secondary network does not percolate, percolation may occur in the cooperative secondary network; or when a noncooperative secondary network percolates, less energy may be used to sustain the same level of connectivity in the cooperative secondary network.

We study the degree distribution of a wireless ad hoc network node sustaining inter-system interference from heterogeneous wireless ad hoc networks. The relationship between the average number of neighbors of a wireless node and the number of retransmissions is clarified under different broadcasting schemes. The degree distribution is then generalized to the multi-channel case. Furthermore, the connection between the degree distribution of a wireless node and the connectivity of the wireless network in percolation sense is established by extending the methods of complex network. Using these tools, the benefit of cooperation among heterogeneous wireless networks is analytically justified in terms of percolation threshold. Supplemented with numerical results, our analysis gives insight into fundamental limits of coexistence of heterogeneous wireless networks, advantages of different broadcasting schemes and benefits of cooperation among heterogeneous networks.

To increase spectrum utilization, cognitive radio allows concurrent secondary and primary transmissions as long as interference to primary users is constrained under a threshold. This research proposes an enhanced opportunistic interference mitigation scheme utilizing both successfully and unsuccessfully decoded primary packets to improve data rate of secondary transmission. Moreover, we propose an analytical model to investigate characteristic changes of the spectrum usage affected by the coexisting secondary transmission in terms of overall spectral efficiency. The interference mitigation scheme can be applied to realistic two-tier femtocell networks to enable robust communication against cross-tier interference thereby obtaining a substantial spectrum reuse gain.

Characterizing the topology and therefore fundamental limits is a must to establish effective end-to-end cognitive radio networking (CRN). However, there lacks complete understanding of the relationship among connectivity, interference, latency and other system parameters of the CRN. To clarify this complication, by employing tools from both percolation theory and stochastic geometry, we thus provide a novel parametrization of underlay secondary ad hoc CRN wherein the secondary network is regarded as an operating point in the phase space. Coexisting with a primary ad hoc network, the secondary network undergoes a phase transition due to avoiding interference to primary receivers, while being interfered by primary transmitters. Furthermore, transmit power allocation of secondary users is represented by a Pareto contour in the phase space, and the impact of interference on connectivity is captured by the latency-to- percolate. Finally, with the cognitive capability of CR, performance improvement of importing an SU- avoidance region around primary receivers is analyzed, and CRNs can be therefore successfully supplied.

In two-tier networks consisting of a macrocell overlaid with femtocells in co-channel deployment and closed-access policy, spatial reuse is achieved at the price of severe cross-tier interference from concurrent transmissions. The lack of direct coordination between the macro and femtocells makes interference control as a challenging issue. Cognitive radio (CR) becomes a promising solution, where femtocells with cognitive information accomplish concurrent transmissions while meeting a per-tier outage constraint. Several interference-aware allocation approaches are proposed to enhance spatial reuse according to cognitive capabilities of femtocell. By employing stochastic geometry model, bounds on the distribution of aggregated interference from two-tier spatial point processes are successfully analyzed. The maximum number of simultaneously transmitting femtocells and overall downlink capacity of two-tier networks meeting a per-tier outage requirement in each approach are theoretically derived. This paper proves that with stronger cognitive capability (i.e., more knowledge interpreted) at femtocell, more spatial reuse gain can be found.

Cognitive radio networks (CRN) may greatly enhance the throughput based on a given bandwidth. CRN has a unique feature, consisting of uni-directional opportunistic wireless links, and packets may be cooperatively relayed from source node to destination node through one or multiple paths, while each path consists of multi-hop opportunistic wireless fading links. Effective end-to-end error control is therefore a must to complete transportation of packets in CRN, though not being touched in open literatures. Traditional ARQ cannot generally function over uni-directional opportunistic links. We consequently develop a novel hybrid ARQ based on amplify-and-forward cooperative relay of CRN to reach the purpose of end-to-end error control at session level. Multiple coded sub- packets are sent through multiple paths, and the destination node decodes these received coded sub- packets from different paths. This HARQ for CRN works even missing some coded sub-packets. By the concept of outage, we also study allocation of coded sub-packets on multiple paths in an information theoretical view and develop a coding rate adaptation scheme for CRN such that end-to-end error control is possible.