arXiv subjects

Sharayu Moharir

Publications and source records attributed to Sharayu Moharir.

At least 37 records · Page 2Linked to original sources

arXiv2021

RetroRenting: An Online Policy for Service Caching at the Edge

The rapid proliferation of shared edge computing platforms has enabled application service providers to deploy a wide variety of services with stringent latency and high bandwidth requirements. A key advantage of these platforms is that they provide pay-as-you-go flexibility by charging clients in proportion to their resource usage through short-term contracts. This affords the client significant cost-saving opportunities, by dynamically deciding when to host its service on the platform, depending on the changing intensity of requests. A natural policy for our setting is the Time-To-Live (TTL) policy. We show that TTL performs poorly both in the adversarial arrival setting, i.e., in terms of the competitive ratio, and for i.i.d. stochastic arrivals with low arrival rates, irrespective of the value of the TTL timer. We propose an online policy called RetroRenting (RR) and show that in the class of deterministic online policies, RR is order-optimal with respect to the competitive ratio. In addition, we provide performance guarantees for RR for i.i.d. stochastic arrival processes coupled with negatively associated rent cost sequences and prove that it compares well with the optimal online policy. Further, we conduct simulations using both synthetic and real world traces to compare the performance of RR with the optimal offline and online policies. The simulations show that the performance of RR is near optimal for all settings considered. Our results illustrate the universality of RR.

cs.NI↗

arXiv2021

Decentralized Age-of-Information Bandits

Age-of-Information (AoI) is a performance metric for scheduling systems that measures the freshness of the data available at the intended destination. AoI is formally defined as the time elapsed since the destination received the recent most update from the source. We consider the problem of scheduling to minimize the cumulative AoI in a multi-source multi-channel setting. Our focus is on the setting where channel statistics are unknown and we model the problem as a distributed multi-armed bandit problem. For an appropriately defined AoI regret metric, we provide analytical performance guarantees of an existing UCB-based policy for the distributed multi-armed bandit problem. In addition, we propose a novel policy based on Thomson Sampling and a hybrid policy that tries to balance the trade-off between the aforementioned policies. Further, we develop AoI-aware variants of these policies in which each source takes its current AoI into account while making decisions. We compare the performance of various policies via simulations.

eess.SY↗

arXiv2021

On Minimizing Channel-Aware Age of Information in a Multi-Sensor Setting

We propose a variant of the Age of Information (AoI) metric called Channel-Aware Age of Information (CA-AoI). Unlike AoI, CA-AoI takes into account the channel conditions between the source and the intended destination to compute the "age" of the recent most update received by the destination. This new metric ensures that the resource allocation is not heavily tilted towards the sources with poor channel conditions. We design scheduling policies for multi-sensor systems in which sensors report their measurements to a central monitoring station via a shared unreliable communication channel with the goal of minimizing the time-average of the weighted sum of CA-AoIs. We initially derive universal lower bounds for the freshness objective. We show that the scheduling problem is indexable and derive low complexity Whittle index based scheduling policies. We also design stationary randomized scheduling algorithms and give optimization procedures to find the optimal parameters of the policy. Via simulations, we show that our proposed policies surpass the greedy policy in several settings. Moreover the Whittle Index based scheduling policies outperform other policies in all the settings considered.

eess.SY↗

arXiv2020

Correlated Age-of-Information Bandits

We consider a system composed of a sensor node tracking a time varying quantity. In every discretized time slot, the node attempts to send an update to a central monitoring station through one of K communication channels. We consider the setting where channel realizations are correlated across channels. This is motivated by mmWave based 5G systems where line-of-sight which is critical for successful communication is common across all frequency channels while the effect of other factors like humidity is frequency dependent. The metric of interest is the Age-of-Information (AoI) which is a measure of the freshness of the data available at the monitoring station. In the setting where channel statistics are unknown but stationary across time and correlated across channels, the algorithmic challenge is to determine which channel to use in each time-slot for communication. We model the problem as a Multi-Armed bandit (MAB) with channels as arms. We characterize the fundamental limits on the performance of any policy. In addition, via analysis and simulations, we characterize the performance of variants of the UCB and Thompson Sampling policies that exploit correlation.

eess.SY↗

arXiv2020

Regret of Age-of-Information Bandits

We consider a system with a single source that measures/tracks a time-varying quantity and periodically attempts to report these measurements to a monitoring station. Each update from the source has to be scheduled on one of K available communication channels. The probability of success of each attempted communication is a function of the channel used. This function is unknown to the scheduler. The metric of interest is the Age-of-Information (AoI), formally defined as the time elapsed since the destination received the recent most update from the source. We model our scheduling problem as a variant of the multi-arm bandit problem with communication channels as arms. We characterize a lower bound on the AoI regret achievable by any policy and characterize the performance of UCB, Thompson Sampling, and their variants. Our analytical results show that UCB and Thompson sampling are order-optimal for AoI bandits. In addition, we propose novel policies which, unlike UCB and Thompson Sampling, use the current AoI to make scheduling decisions. Via simulations, we show the proposed AoI-aware policies outperform existing AoI-agnostic policies.

eess.SY↗

arXiv2020

Coded Caching for Heterogeneous Wireless Networks

This chapter provides an overview of coded caching in the context of heterogeneous wireless networks. We begin by briefly describing the key idea behind coded caching and then discuss in detail the impact of various aspects such as non-uniform content popularity, multiple cache access, and interference.

cs.IT↗

arXiv2020

Influencing Opinions of Heterogeneous Populations over Finite Time Horizons

In this work, we focus on strategies to influence the opinion dynamics of a well-connected society. We propose a generalization of the popular voter model. This variant of the voter model can capture a wide range of individuals including strong-willed individuals whose opinion evolution is independent of their neighbors as well as conformist/rebel individuals who tend to adopt the opinion of the majority/minority. Motivated by political campaigns which aim to influence opinion dynamics by the end of a fixed deadline, we focus on influencing strategies for finite time horizons. We characterize the nature of optimal influencing strategies as a function of the nature of individuals forming the society. Using this, we show that for a society consisting of predominantly strong-willed/rebel individuals, the optimal strategy is to influence towards the end of the finite time horizon, whereas, for a society predominantly consisting of conformist individuals who try to adopt the opinion of the majority, it could be optimal to influence in the initial phase of the finite time horizon.

cs.SI↗

arXiv2020

Influencing Opinion Dynamics in Networks with Limited Interaction

The focus of this work is on designing influencing strategies to shape the collective opinion of a network of individuals. We consider a variant of the voter model where opinions evolve in one of two ways. In the absence of external influence, opinions evolve via interactions between individuals in the network, while, in the presence of external influence, opinions shift in the direction preferred by the influencer. We focus on a finite time-horizon and an influencing strategy is characterized by when it exerts influence in this time-horizon given its budget constraints. Prior work on this opinion dynamics model assumes that individuals take into account the opinion of all individuals in the network. We generalize this and consider the setting where the opinion evolution of an individual depends on a limited collection of opinions from the network. We characterize the nature of optimal influencing strategies as a function of the way in which this collection of opinions is formed.

cs.SI↗

arXiv2020

Duopolistic platform competition for revenue and throughput

We consider two competing platforms operating in a two-sided market and offering identical services to their customers at potentially different prices. The objective of each platform is to maximize its throughput or revenue by suitably pricing its services. We assume that customers have a preference or loyalty to the platforms while the workers freelance for the two platforms. Assuming that the resulting interaction between the users is such that their aggregate utility is maximized, we show that for each value of the loyalty, there exists a pure strategy Nash equilibrium for both the throughput and revenue competition game and characterize it.

math.OC↗

arXiv2020

Resource Pooling in Large-Scale Content Delivery Systems

Content delivery networks are a key infrastructure component used by Video on Demand (VoD) services to deliver content over the Internet. We study a content delivery system consisting of a central server and multiple co-located caches, each with limited storage and service capabilities. This work evaluates the performance of such a system as a function of the storage capacity of the caches, the content replication strategy, and the service policy. This analysis can be used for a system-level optimization of these design choices. The focus of this work is on understanding the benefits of allowing caches to pool their resources to serve user requests. We show that the benefits of resource pooling depend on the popularity profile of the contents offered by the VoD service. More specifically, if the popularity does not vary drastically across contents, resource pooling can lead to significant improvements in the system performance. In contrast, if the content popularity is uneven, the benefits of resource pooling are negligible.

cs.IT↗

arXiv2019

Contextual Bandits Evolving Over Finite Time

Contextual bandits have the same exploration-exploitation trade-off as standard multi-armed bandits. On adding positive externalities that decay with time, this problem becomes much more difficult as wrong decisions at the start are hard to recover from. We explore existing policies in this setting and highlight their biases towards the inherent reward matrix. We propose a rejection based policy that achieves a low regret irrespective of the structure of the reward probability matrix.

cs.LG↗

arXiv2019

Unreliable Multi-Armed Bandits: A Novel Approach to Recommendation Systems

We use a novel modification of Multi-Armed Bandits to create a new model for recommendation systems. We model the recommendation system as a bandit seeking to maximize reward by pulling on arms with unknown rewards. The catch however is that this bandit can only access these arms through an unreliable intermediate that has some level of autonomy while choosing its arms. For example, in a streaming website the user has a lot of autonomy while choosing content they want to watch. The streaming sites can use targeted advertising as a means to bias opinions of these users. Here the streaming site is the bandit aiming to maximize reward and the user is the unreliable intermediate. We model the intermediate as accessing states via a Markov chain. The bandit is allowed to perturb this Markov chain. We prove fundamental theorems for this setting after which we show a close-to-optimal Explore-Commit algorithm.

stat.ML↗

arXiv2018

Effects of Storage Heterogeneity in Distributed Cache Systems

In this work, we focus on distributed cache systems with non-uniform storage capacity across caches. We compare the performance of our system with the performance of a system with the same cumulative storage distributed evenly across the caches. We characterize the extent to which the performance of the distributed cache system deteriorates due to storage heterogeneity. The key takeaway from this work is that the effects of heterogeneity in the storage capabilities depend heavily on the popularity profile of the contents being cached and delivered. We analytically show that compared to the case where contents popularity is comparable across contents, lopsided popularity profiles are more tolerant to heterogeneity in storage capabilities. We validate our theoretical results via simulations.

cs.IT↗

arXiv2018

Caching with Partial Adaptive Matching

We study the caching problem when we are allowed to match each user to one of a subset of caches after its request is revealed. We focus on non-uniformly popular content, specifically when the file popularities obey a Zipf distribution. We study two extremal schemes, one focusing on coded server transmissions while ignoring matching capabilities, and the other focusing on adaptive matching while ignoring potential coding opportunities. We derive the rates achieved by these schemes and characterize the regimes in which one outperforms the other. We also compare them to information-theoretic outer bounds, and finally propose a hybrid scheme that generalizes ideas from the two schemes and performs at least as well as either of them in most memory regimes.

cs.IT↗

arXiv2017

Caching under Content Freshness Constraints

Several real-time delay-sensitive applications pose varying degrees of freshness demands on the requested content. The performance of cache replacement policies that are agnostic to these demands is likely to be sub-optimal. Motivated by this concern, in this paper, we study caching policies under a request arrival process which incorporates user freshness demands. We consider the performance metric to be the steady-state cache hit probability. We first provide a universal upper bound on the performance of any caching policy. We then analytically obtain the content-wise hit-rates for the Least Popular (LP) policy and provide sufficient conditions for the asymptotic optimality of cache performance under this policy. Next, we obtain an accurate approximation for the LRU hit-rates in the regime of large content population. To this end, we map the characteristic time of a content in the LRU policy to the classical Coupon Collector's Problem and show that it sharply concentrates around its mean. Further, we develop modified versions of these policies which eject cache redundancies present in the form of stale contents. Finally, we propose a new policy which outperforms the above policies by explicitly using freshness specifications of user requests to prioritize among the cached contents. We corroborate our analytical insights with extensive simulations.

cs.NI↗

arXiv2017

Request Patterns and Caching for VoD Services with Recommendation Systems

Video on Demand (VoD) services like Netflix and YouTube account for ever increasing fractions of Internet traffic. It is estimated that this fraction will cross 80% in the next three years. Most popular VoD services have recommendation engines which recommend videos to users based on their viewing history, thus introducing time-correlation in user requests. Understanding and modeling this time-correlation in user requests is critical for network traffic engineering. The primary goal of this work is to use empirically observed properties of user requests to model the effect of recommendation engines on the request patterns in VoD services. We propose a Markovian request model to capture the time-correlation in user requests and show that our model is consistent with the observations of existing empirical studies. Most large-scale VoD services deliver content to users via a distributed network of servers as serving users requests via geographically co-located servers reduces latency and network bandwidth consumption. The content replication policy, i.e., determining which contents to cache on the servers is a key resource allocation problem for VoD services. Recent studies show that low start-up delay is a key Quality of Service (QoS) requirement of users of VoD services. This motivates the need to pre-fetch (fetch before contents are requested) and cache content likely to be request in the near future. Since pre-fetching leads to an increase in the network bandwidth usage, we use our Markovian model to explore the trade-offs and feasibility of implementing recommendation based pre-fetching.

cs.NI↗

arXiv2016

Content Replication in Large Distributed Caches

In this paper, we consider the algorithmic task of content replication and request routing in a distributed caching system consisting of a central server and a large number of caches, each with limited storage and service capabilities. We study a time-slotted system where in each time-slot, a large batch of requests has to be matched to a large number of caches, where each request can be served by any cache which stores the requested content. All requests which cannot be served by the caches are served by fetching the requested content from the central server. The goal is to minimize the transmission rate from the central server. We use a novel mapping between our content replication problem and the Knapsack problem to prove a lower bound on the transmission rate for any content replication policy. Using insights obtained from the mapping, we propose a content replication policy - Knapsack Storage - which achieves this lower bound. While it intuitively makes sense to replicate the more popular contents on a larger number of caches, surprisingly, in certain cases, the Knapsack Storage policy chooses not to replicate the most popular contents on the caches at all.

cs.NI↗

arXiv2016

Paging with Multiple Caches

Modern content delivery networks consist of one or more back-end servers which store the entire content catalog, assisted by multiple front-end servers with limited storage and service capacities located near the end-users. Appropriate replication of content on the front-end servers is key to maximize the fraction of requests served by the front-end servers. Motivated by this, a multiple cache variant of the classical single cache paging problem is studied, which is referred to as the Multiple Cache Paging (MCP) problem. In each time-slot, a batch of content requests arrive that have to be served by a bank of caches, and each cache can serve exactly one request. If a content is not found in the bank, it is fetched from the back-end server, and one currently stored content is ejected, and counted as fault. As in the classical paging problem, the goal is to minimize the total number of faults. The competitive ratio of any online algorithm for the MCP problem is shown to be unbounded for arbitrary input, thus concluding that the MCP problem is fundamentally different from the classical paging problem. Consequently, stochastic arrivals setting is considered, where requests arrive according to a known/unknown stochastic process. It is shown that near optimal performance can be achieved with simple policies that require no co-ordination across the caches.

cs.DS↗