Avik Sengupta

Cache aided wireless networks: Tradeoffs between storage and latency

We investigate the fundamental information theoretic limits of cache-aided wireless networks, where edge nodes (or transmitters) are endowed with caches that can store popular content such as multimedia files. This architecture aims to localize popular multimedia content by proactively pushing it closer to the edge of the wireless network, thereby alleviating backhaul load. An information theoretic model of such networks is presented, that includes the introduction of a new metric, namely normalized delivery time (NDT), which captures the worst case time to deliver any requested content to the users. We present new results on the trade-off between latency, measured via the NDT, and the cache storage capacity of the edge nodes. In particular, a novel information theoretic lower bound on NDT is presented for cache aided networks. The optimality of this bound is shown for several system parameters.

Fog-Aided Wireless Networks for Content Delivery: Fundamental Latency Tradeoffs

A fog-aided wireless network architecture is studied in which edge nodes (ENs), such as base stations, are connected to a cloud processor via dedicated fronthaul links while also being endowed with caches. Cloud processing enables the centralized implementation of cooperative transmission strategies at the ENs, albeit at the cost of an increased latency due to fronthaul transfer. In contrast, the proactive caching of popular content at the ENs allows for the low-latency delivery of the cached files, but with generally limited opportunities for cooperative transmission among the ENs. The interplay between cloud processing and edge caching is addressed from an information-theoretic viewpoint by investigating the fundamental limits of a high signal-to-noise-ratio metric, termed normalized delivery time (NDT), which captures the worst case coding latency for delivering any requested content to the users. The NDT is defined under the assumptions of either serial or pipelined fronthaul-edge transmission, and is studied as a function of fronthaul and cache capacity constraints. Placement and delivery strategies across both fronthaul and wireless, or edge, segments are proposed with the aim of minimizing the NDT. Information-theoretic lower bounds on the NDT are also derived. Achievability arguments and lower bounds are leveraged to characterize the minimal NDT in a number of important special cases, including systems with no caching capabilities, as well as to prove that the proposed schemes achieve optimality within a constant multiplicative factor of 2 for all values of the problem parameters.

Cloud RAN and edge caching: Fundamental performance trade-offs

A wireless network architecture is studied in which edge nodes (ENs), such as base stations, are connected to a cloud processor by dedicated front haul links, while also being endowed with caches, in which popular content, such as multimedia files, can be proactively stored. Cloud processing enables the centralized implementation of cooperative transmission by the ENs, albeit at the cost of an increased latency due to fronthaul transfer. In contrast, edge caching allows for the low-latency delivery of the cached files, but with generally limited cooperation among the ENs. The interplay between cloud processing and edge caching is studied from an information-theoretic viewpoint by investigating the fundamental limits of a metric, termed normalized delivery time (NDT), which captures the worst-case latency for delivering any requested content to the users. Lower and upper bounds on the NDT are derived that yield insights into the trade-off between cache storage capacity, fronthaul capacity and delivery latency.

On the performance of redundant residue number system codes assisted STBC design

In this paper, we propose a novel application of Redundant Residue Number System (RRNS) codes to Space-Time Block Codes (STBCs) design. Based on the so-called “Direct-Mapping” scheme, the link between residues and complex signal constellations is optimized. We derive upper bounds on the codeword error probability of RRNS-STBC and characterize its achievable diversity gain assuming maximum likelihood decoding (MLD). The knowledge of apriori probabilities of residue generation is utilized to implement a probability based Distance-Aware Direct Mapping scheme for M-ary modulation which further improves the error performance of the RRNS-STBC coding scheme.

Online edge caching in fog-aided wireless networks

In a Fog Radio Access Network (F-RAN) architecture, edge nodes (ENs), such as base stations, are equipped with limited-capacity caches, as well as with fronthaul links that can support given transmission rates from a cloud processor. Existing information-theoretic analyses of content delivery in F-RANs have focused on offline caching with separate content placement and delivery phases. In contrast, this work considers an online caching set-up, in which the set of popular files is time-varying and both cache replenishment and content delivery can take place in each time slot. The analysis is centered on the characterization of the long-term Normalized Delivery Time (NDT), which captures the temporal dependence of the coding latencies accrued across multiple time slots in the high signal-to-noise ratio regime. Online caching and delivery schemes based on reactive and proactive caching are investigated, and their performance is compared to optimal offline caching schemes both analytically and via numerical results.

Improved Approximation of Storage-Rate Tradeoff for Caching With Multiple Demands

Caching at the network edge has emerged as a viable solution for alleviating the severe capacity crunch in modern content centric wireless networks by leveraging network load-balancing in the form of localized content storage and delivery. In this paper, we consider a cache-aided network, where the cache storage phase is assisted by a central server and users can demand multiple files at each transmission interval. To service these demands, we consider two delivery models: 1) centralized content delivery, where user demands at each transmission interval are serviced by the central server via multicast transmissions; and 2) device-to-device assisted distributed delivery, where users multicast to each other in order to service file demands. For such cache-aided networks, we present new results on the fundamental cache storage versus transmission rate tradeoff. Specifically, we develop a new technique for characterizing information theoretic lower bounds on the storage-rate tradeoff and show that the new lower bounds are strictly tighter than cut-set bounds from literature. Furthermore, using the new lower bounds, we improve the constant factor approximation of the optimal storage-rate tradeoff for cache-aided systems under both delivery models.

Beyond cut-set bounds - the approximate capacity of D2D networks

Device-to-Device (D2D) communication is emerging as a viable solution for alleviating the severe capacity crunch in content-centric wireless networks. D2D encourages backhaul-free communication directly between devices with similar content requirements grouped into clusters. In this work, a self-sustaining D2D network is considered, where a set of commonly requested files are completely stored within the collective devices memories in a cluster and file requests from devices are serviced by local inter-device multicast transmissions. For such a network, new information theoretic converse results are developed, in the form of a lower bound on the minimum D2D multicast rate as a function of the storage per device. The proposed converse is then used to characterize the approximate tradeoff between the device storage and D2D multicast rate to within a constant multiplicative gap of 8.

Pipelined Fronthaul-Edge Content Delivery in Fog Radio Access Networks

In a Fog Radio Access Network (F-RAN), content delivery is carried out using both edge caching and cloud processing. A key design question for F-RANs hence concerns the optimal use of edge and cloud resources. In this work, this problem is addressed from an information theoretic viewpoint by investigating the fundamental limits of the normalized delivery time (NDT) metric, which captures the high signal- to-noise ratio (SNR) worst-case latency for delivering any requested content to the users. Specifically, unlike prior work, the NDT perfor- mance of an F-RAN is studied under pipelined fronthaul-edge transmission, whereby edge nodes are capable of simultaneously receiving fronthaul messages from the cloud on fronthaul links while transmitting to the mobile users over the wireless edge channel. Lower and upper bounds on the NDT are derived that yield insights into the trade-off between cache storage capacity, fronthaul capacity and delivery latency and on the impact of fronthaul-edge pipelining.

Layered caching for heterogeneous storage

In modern data-centric wireless networks, caching alleviates severe capacity crunch at times of high network load. Recent results have shown that careful design of cache storage to leverage coded multicast file delivery over a shared link can achieve order-wise improvements in the storage-rate trade-off. In this work, we present a novel caching and delivery scheme for the case when users have heterogeneous cache sizes. The proposed scheme uses two new ingredients namely set partitioning and cache layering. The main challenge in designing caching schemes in presence of storage heterogeneity is that varying levels of storage across users can present a variety of caching and multicasting opportunities. Our framework of cache layering and set partitioning is a principled approach to utilize such opportunities, where each layer delivers a fraction of requested data to a specific set of users and layers operate independently of each other. We also derive an information-theoretic lower bound for the heterogeneous caching problem.

Malware propagation in fully connected networks: A netflow-based analysis

Malware attacks have become ubiquitous in modern large data-centric networks. Therefore advanced malware threat detection and related countermeasures are an important paradigm in cybersecurity research. This work studies malware propagation in fully connected networks, where network topology plays a minimal role in lateral spread within the network. The live netflow and perimeter alert data used in this study contrasts with other previous works due to the unavailability of ground truth for any attack type. Important features calculated from the netflow data as well as a novel ring-based flow model are described. These are helpful in tracking possible malware flow within the network. The results show that relevant features can be used to draw inferences about the propagation of certain classes of malware attacks.