Esa Hyytia

• Dr.
• Professor (Associate) at University of Iceland

The problem of routing jobs to parallel servers is known as the dispatching problem. A typical objective is to minimize the mean response time, which according to Little’s result is equivalent to minimizing the mean number in the system. Dynamic dispatching policies are based on information about the state of each server. In large or real-time syst...

Parallel server systems are ubiquitous. Multicore CPUs are in practically every personal device from mobile handsets to high-end desktop PCs. At larger scale, data centers consist of a huge number of physical servers often shared by multiple users (for economic reasons). Moreover, the simultaneous users are typically unaware of each other due to re...

We develop a unified framework for analyzing and optimizing costs for systems of FCFS queues with batch arrivals, setup delays and a general nonlinear cost structure that includes costs associated with energy used, setup times and Quality of Service (QoS) measures. We focus on the MX/G/1 and GeoX/G/1 queues with i.i.d. service times, but our result...

Many businesses possess a small infrastructure that they can use for their computing tasks, but also often buy extra computing resources from clouds. Cloud vendors such as Amazon EC2 offer two types of purchase options: on-demand and spot instances. As tenants have limited budgets to satisfy their computing needs, it is crucial for them to determin...

Dispatching systems, where jobs are routed to servers immediately upon arrival, appear frequently in parallel computing systems of different scales. With a dynamic dispatching policy, the system is generally analytically intractable and performance evaluation is carried out by means of Monte Carlo simulations. A typical performance metric is the me...

Redundancy is an increasingly popular technique for reducing response times in computer systems, and there is a growing body of theoretical work seeking to analyze performance in systems with redundancy. The idea is to dispatch a job to multiple servers at the same time and wait for the first copy to complete service. Redundancy can help reduce res...

We study the M/D/1 queue and its generalization, the M/iD/1 queue, when jobs have firm deadlines for waiting (or sojourn) time. If a deadline is not met, a job-specific deadline violation cost is incurred. The M/iD/1 queue enables us to model both varying jobs sizes and batch arrival processes. We derive explicit value functions for these M/D/1-typ...

Routing jobs to parallel servers is a common and important task in today’s computer and communication systems. As each routing decision affects the jobs arriving later, determining the (near) optimal decisions is non-trivial. In this paper, we apply reinforcement learning techniques to the job routing problem with heterogeneous servers and a genera...

Many of the existing opportunistic networking systems have been designed assuming a small number links per node and have trouble scaling to large numbers of potential concurrent communication partners. In the real world we often find wireless local area networks with large numbers of connected users – in particular in open Wi-Fi networks provided b...

Routing jobs to parallel servers is a common and important task in today's computer systems. Join-the-shortest-queue (JSQ) routing minimizes the mean response time under rather general settings as long as the servers are identical and service times are independent and exponentially distributed. Apart from this, surprisingly few optimality results e...

Many businesses possess a small infrastructure that they can use for their computing tasks, but also often buy extra computing resources from clouds. Cloud vendors such as Amazon EC2 offer two types of purchase options: on-demand and spot instances. As tenants have limited budgets to satisfy their computing needs, it is crucial for them to determin...

Server-side variability – the idea that the same job can take longer to run on one server than another due to server-dependent factors – isan increasingly important concern in many queueing systems. One strategy for overcoming server-side variability to achieve low response time is redundancy, under which jobs create copies of themselves and send t...

We study the M/D/1 queue when jobs have firm deadlines for waiting (or sojourn) time. If a deadline is not met, a job-specific deadline violation cost is incurred. We derive explicit value functions for this M/D/1 queue that enable the development of efficient cost-aware dispatching policies to parallel servers. The performance of the resulting dis...

A number of opportunistic content sharing services were developed that exploit device-to-device contacts for infrastructure-less operation. All of them depend, like geo-based ad-hoc routing protocols, on mobile devices knowing their respective positions to accurately perform data replication. In this paper, we explore the impact of different types...

Many of the existing opportunistic networking systems have been designed assuming a small number links per node and have trouble scaling to large numbers of potential concurrent communication partners. In the real world we often find wireless local area networks with large numbers of connected users -- in particular in open Wi-Fi networks provided...

Recent computer systems research has proposed using redundant requests to reduce latency. The idea is to run a request on multiple servers and wait for the first completion (discarding all remaining copies of the request). However, there is no exact analysis of systems with redundancy. This paper presents the first exact analysis of systems with re...

Cloud vendors often offer three types of instances: reserved, on-demand and spot instances; a premier challenge for all users is to determine the proportion of various instances so as to minimize the cost of completing all jobs while satisfying their response-time targets. A previous online learning framework enables dynamically acquiring on-demand...

We consider a dispatching system, where jobs with deadlines for their waiting times are assigned to FCFS servers immediately upon arrival. The dispatching problem is to choose a server for each job so as to minimize the probability of deadline violation. We derive an efficient deadline-aware policy in the MDP framework by means of policy improvemen...

Due to the fierce competition for the wireless spectrum, operators have recently focused on short-range communications, which promises higher spectral efficiency, lower energy consumption, and less strain on the operators¿¿¿ core networks. Mobile opportunistic communications, that is, short-range communications without any network assistance, is of...

We study the Round-Robin (RR) routing to a system of parallel queues, which serve jobs according to FCFS or preemptive LCFS scheduling disciplines. The cost structure comprises two components: a service fee and a queueing delay related component. With Poisson arrivals, the inter-arrival time to each queue obeys the Erlang distribution. This allows...

While there is a drastic shift from host-centric networking to content-centric networking, how to locate and retrieve the relevant content efficiently, especially in a mobile network, is still an open question. Mobile devices host increasing volume of data which could be shared with the nearby nodes in a multi-hop fashion. However, searching for co...

Searching content in mobile opportunistic networks is a difficult problem due to the dynamically changing topology and intermittent connections. Moreover, due to the lack of global view of the network, it is arduous to determine whether the best response is discovered or search should be spread to other nodes. A node that has received a search quer...

We consider a dynamic offloading problem arising in the context of mobile cloud computing (MCC). In MCC, three types of tasks can be identified: (i) those which can be processed only locally in a mobile device, (ii) those which are processed in the cloud, and (iii) those which can be processed either in the mobile or in the cloud. For type (iii) ta...

Recent computer systems research has proposed using redundant requests to reduce latency. The idea is to run a request on multiple servers and wait for the first completion (discarding all remaining copies of the request). However there is no exact analysis of systems with redundancy. This paper presents the first exact analysis of systems with red...

We consider the job (or task) assignment problem to heterogeneous parallel servers, where servers can be switched off to save energy. However, switching a server back on involves a constant server-specific delay. We will use one step of policy iteration from a starting policy such as Bernoulli splitting, in order to derive efficient job assignment...

Searching content in mobile opportunistic networks is a difficult problem due to the dynamically changing topology and intermittent connections. Moreover, due to the lack of global view of the network, it is arduous to determine whether the best response is discovered or search should be spread to other nodes. A node that has received a search quer...

To date, the study of dispatching or load balancing in server farms has primarily focused on the minimization of response time. Server farms are typically modeled by a front-end router that employs a dispatching policy to route jobs to one of several servers, with each server scheduling all the jobs in its queue via Processor-Sharing. However, the...

We consider the task assignment problem to heterogeneous parallel servers with switching delay, where servers can be switched off to save energy. However, switching a server back on involves a constant server-specific delay. We will use one step of policy iteration from a starting policy such as Bernoulli splitting, in order to derive efficient tas...

We study the Round-Robin (RR) routing to a system of parallel queues. The cost structure comprises two components: a service fee and a queueing delay related component, where both can be job- and queue-specific random variables. With Poisson arrivals, the inter-arrival time to each queue obeys Erlang's distribution. This allows us to study the mean...

To detect peers in mobile opportunistic networks, mobile devices transmit and listen for beacons (“scanning”). If networks are sparse, devices spend quite a bit of energy scanning the vicinity for possible contacts with their radios. Numerous techniques were developed to adapt the scanning intervals as a function of the observed node density. In th...

The Opportunistic Networking Environment (ONE) Simulator is an extensible tool for evaluating protocols and mobility models for delay-tolerant networking. ONE allows easily plugging in mobility models, contact traces, routing modules, applications, and report modules. In this paper, we describe how to instrument the ONE simulator for two content sh...

We consider a heterogeneous two-server system processing fixed size jobs. This includes the scheduling system, where jobs wait in a common queue, and the dispatching system, where jobs are assigned to server-specific queues upon arrival. The optimal policy with respect to the delay in both systems is a threshold policy characterized by a single par...

Applying the first policy iteration (FPI) to any static dispatching (task assignment) policy yields a new improved dynamic policy that takes into account the particular cost structure and the expected future arrivals. However, it is generally hard to go beyond that due to the complex state space and the resulting difficulty in computing the value f...

The increase in the size of mobile cloud as well as the volume of information necessitates efficient search mechanisms for finding the searched information or the target node. In this paper, we focus on search mechanisms to retrieve information from within a mobile cloud in which nodes have intermittent connectivity and hence operate on a store-car...

We consider a dispatching system, where jobs, arriving in batches, are assigned to single-server FCFS queues. Batches can be split to different queues on per job basis. However, the holding costs are batch-specific and incurred until the last member of the batch completes the service. By using the first policy improvement step of the MDP framework,...

For mobile delay-tolerant networks, different mobility models have been utilized to assess the performance of routing algorithms and applications. Substantial work has gone into understanding the contact characteristics of mobile users to allow evaluation under conditions that approximate the real world. One important finding has been recognizing t...

In an opportunistic content sharing system referred to as floating content, information is copied between mobile nodes upon node encounters inside an area which is called the anchor zone. We study the conditions under which information can be sustained in such a system. The anchor zone is assumed to be a circular disk, and a random walk type mobili...

We consider a single-server queue with Poisson input operating under first-come--first-served (FCFS) or last-come--first-served (LCFS) disciplines. The service times of the customers are independent and obey a general distribution. The system is subject to costs for holding a customer per unit of time, which can be customer specific or customer cla...

In dial-a-ride problems, a Beet of n vehicles is routed to transport people between pick-up and delivery locations. We consider an elementary version of the problem where trip requests arrive in time and require an immediate vehicle assignment (which triggers an appropriate route update of the selected vehicle). In this context, a relatively genera...

We study delay tolerant networking (DTN) and in particular, its capacity to store, carry and forward messages so that the messages eventually reach their final destinations. We approach this broad question in the framework of percolation theory. To this end, we assume an elementary mobility model, where nodes arrive to an infinite plane according t...

We consider the critical percolation threshold for aligned cylinders, which provides a lower bound for the required node degree for the permanence of information in opportunistic networking. The height of a cylinder corresponds to the time a node is active in its current location. By means of Monte Carlo simulations, we obtain an accurate numerical...

We consider a system of parallel queues where tasks are assigned (dispatched) to one of the available servers upon arrival. The dispatching decision is based on the full state information, i.e., on the sizes of the new and existing jobs. We are interested in minimizing the so-called mean slowdown criterion corresponding to the mean of the sojourn t...

In the Floating Content application, mobile nodes collectively store and disseminate messages relevant to a certain area by using the principles of opportunistic networking. The system operates in best effort fashion relying solely on the nodes located in the area of interest, which is referred to as the anchor zone of the message. Past work has fo...

We consider a system of parallel queues where tasks are assigned (dispatched) to one of the available servers upon arrival. The dispatching decision is based on the full state information, i.e., on the sizes of the new and existing jobs. We are interested in minimizing the so-called mean slowdown criterion corresponding to the mean of the sojourn t...

We consider the dispatching problem in a size- and state-aware multi-queue system with Poisson arrivals and queue-specific job sizes. By size- and state-awareness, we mean that the dispatcher knows the size of an arriving job and the remaining service times of the jobs in each queue. By queue-specific job sizes, we mean that the time to process a j...

People are increasingly using online social networks for maintaining contact with friends and colleagues irrespective of their physical location. While such services are essential to overcome distances, using infrastructure services for location-based services may not be desirable. In contrast, we design and analyze a fully distributed variant of a...

We consider a distributed server system where heterogeneous servers operate under the processor sharing (PS) discipline. Exponentially distributed jobs arrive to a dispatcher, which assigns each task to one of the servers. In the so-called size-aware system, the dispatcher is assumed to know the remaining service requirements of the existing jobs i...

We consider a dynamic dispatching problem where jobs are assigned upon arrival into parallel queues. Jobs have an arbitrary size distribution and each queue has its own service rate, queueing discipline, and operating power while serving jobs. Our goal is to minimize a weighted sum of delay and energy consumption under the assumption that the dispa...

We consider a distributed server system with m servers operating under the processor sharing (PS) discipline. A stream of fixed size tasks arrives to a dispatcher, which assigns each task to one of the servers. We are interested in minimizing the mean sojourn time, i.e., the mean response time. To this end, we first analyze an M/D/l-PS queue in the...

Delay-tolerant and opportunistic networking are widely investigated for information exchange in (sparse) mobile scenarios, in the eval-uation of which numerous mobility models and traces have been employed. These networking techniques may also be applied in static and fairly dense scenarios for non-directed information shar-ing, in which nodes may...

In this paper, we argue that a large-scale demand responsive system is the missing element from the spectrum of the urban transport modes. To this end, we provide a comprehensive analysis addressing this broad question from several aspects including the customers' perception, various technical aspects, and the related societal aspects. In each case...

We consider an opportunistic content sharing system designed to store and distribute local spatio-temporal “floating” information in uncoordinated P2P fashion relying solely on the mobile nodes passing through the area of interest, referred to as the anchor zone. Nodes within the anchor zone exchange the information in opportunistic manner, i.e., w...

Content sharing using personal web pages, blogs, or online social networks is a common means for people to maintain contact with their friends, colleagues, and acquaintances. While such means are essential to overcome distances, using infrastructure services for location-based services may not be desirable. In this paper, we analyze a fully distrib...

We consider a stochastic model for a system which can serve n customers concurrently, and each accepted and departing customer generates a service interruption. The proposed model describes a single vehicle in a dial-a-ride transport system and is closely related to Erlang’s loss system. We give closed-form expressions for the blocking probability,...

In a dynamic dial-a-ride problem (DARP) the task is to provide a transportation service in a given area by dynamically routing a set of vehicles in response to passengers’ trip requests. Passengers share vehicles similarly as with buses, while the schedule and routes are chosen ad hoc. Each trip is defined by the origin-destination pair in plane au...

Requirements for a traffic monitoring system can be very demanding as both privacy and performance aspects have to be taken into account jointly. Moreover, the legislation sets forth strict rules that must also be met. Various cryptographic primitives provide invaluable tools for realising privacy enforcing mechanisms in such a system with respect...

We study a variant of dynamic vehicle routing problem with pickups and deliveries where a vehicle is allocated to each service (i.e., trip) request immediately upon the arrival of the request. Solutions to this problem can be character-ized as dynamic policies that define how each customer is handled by operating a fleet of vehicles. Evaluation of...

Future networks will be characterized by a dramatic increase in terms of capacity, traffic volume, number of delivered flows, and diffusion capillarity. The emerging concern is how to avoid that a massive gathering of measurement data, and the possible misuse of such collected data, may become a threat to the network customers privacy. To face this...

Many of the cryptographic primitives can be used in several ways. One interesting application of the Shamir's secret sharing scheme in the context of privacy aware traffic monitoring is to escrow a secret key after m suspicious events have been observed [1]. In the proposed system, a so-called front-end component encrypts the monitored data traffic...

This paper introduces a new method to detect and track Skype traffic and users by exploiting cross layer information available within 3G mobile cellular networks. In a 3G core network all flows can be analyzed on a per user basis. A detected Skype message is therefore related to a specific user. This information enables user profiles that provide a...

A B S T R A C T We consider the load balancing problem in large wireless multi-hop networks by applying the continuum approximation. The task is to find routes, geometric curves, such that the maximal traffic load in the net- work is minimized. In finite fixed networks, multi-path route s generally yield a lower congestion and thus allow higher thr...

We consider the load balancing problem in wireless multi-hop networks. In the limit of a dense network, there is a strong separation between the macroscopic and microscopic scales, and the load balancing problem can be formulated as finding continuous curves ("routes") between all source-destination pairs that minimize the maximum of the so-called...

We consider the load balancing problem in large wireless multi-hop networks, often referred to as massively dense wireless multi-hop networks. A network is considered to be massively dense if there are nodes practically everywhere and a typical distance between two nodes is much larger than the transmission range necessitating communication over a...

Fountain codes provide an efficient way to transfer information over erasure channels. We give an exact performance analysis of a specific type of fountain codes, called LT codes, when the message length N is small. Two different approaches are developed. In a Markov chain approach the state space explosion, even with reduction based on permutation...

We study the load balancing problem in a dense wireless multihop network, where a typical path consists of a large number of hops, that is, the spatial scales of a typical distance between source and destination and mean distance between the neighboring nodes are strongly separated. In this limit, we present a general framework for analyzing the tr...

In this paper we study the so-called random,waypoint (RWP) mobility model in the context of cellular networks. In the RWP model the nodes, i.e., mobile users, move along a zigzag path consisting of straight legs from one waypoint to the next. Each waypoint is assumed,to be drawn from the uniform distribution over the given convex domain. In this pa...

We study the connectivity properties of an ad hoc network consisting of n nodes each moving according to the Random Waypoint mobility model. In particular, we focus on estimating two quantities, the probability that the network is connected, and the mean durations of the connectivity periods. The accuracy of the approximations is compared against n...

In this paper we introduce a Markovian random waypoint model which allows us to create diverse mobility patterns in the given movement domain. The model allows adjusting random pause times and the momentary velocity. Furthermore, the distribution used to pick the next waypoint may depend on the current location, which allows, e.g., creation of typi...

The random waypoint model (RWP) is one of the most widely used mobility models in performance analysis of ad hoc networks. We analyze the stationary spatial distribution of a node moving according to the RWP model in a given convex area. For this, we give an explicit expression, which is in the form of a one-dimensional integral giving the density...

We consider a simplified model for the rate control of TCP sources. In particular, we assume idealized negative feedbacks upon reaching a certain total sending rate, i.e., at the moment when the total sending rate attains a given capacity limit c one of the TCP sources is given a negative feedback and the source reduces its sending rate in a multip...

We study the load balancing problem in a dense wireless multihop network, where a typical path consists of large number of hops, i.e., the spatial scales of a typical distance between source and destination, and mean distance between the neighboring nodes are strongly separated. In this limit, we present a general framework for analyzing the traffi...

The traffic matrix estimation based on the measurements at the certain points of a fixed network poses an interesting problem, which has also been studied extensively in the literature. In this paper, we consider a similar problem in the setting of dense multihop wireless network. In particular, we assume a large number of nodes with multihop route...

The congestion control algorithm used in Transmission Control Protocol (TCP) is a so-called additive increase multiplicative decrease (AIMD) algorithm. The algorithm and its performance have been studied extensively in the literature. Today, several versions of t he TCP exist, which behave slightly differently when congestion occurs. However, for t...

The random waypoint model (RWP) is one of the most widely used mobility models in per- formance analysis of mobile wireless networks. In this paper we extend the previous work by deriving an analytical formula for the stationary distribution of a node moving according to a RWP model in n-dimensional space.

In wireless multihop networks each node acts as a relay for the other nodes. Consequently, the distribution of the traffic load has a strong spatial dependence. We consider a dense multihop net- work where the routes are approximately straight line segments. To this end we introduce the so-called line segment traversing pro- cess which defines the...

In this paper a generalized (static) routing and wavelength assignment problem is for- mulated, where the objective is to establish a given set of connections into the network while minimizing the number of wavelength channels at the same time. Connections can be normal lightpaths as well as anycast or multicast connections. Furthermore, each reque...

In this paper we analyse previously proposed MAC proto- cols for optical ring networks under static traffic conditions. The analytical models are developed for random order and round-robin transmission policies in both slotted and unslot- ted cases. The models predict the receiver efficiency, i.e. the fraction of time each receiver is active. The m...

In This work we present a linear programming formulation for the routing problem in optical burst switching networks (OBS). Unlike in conventional wavelength routed networks, in an OBS network (light)paths are allowed to clash. One should, however, try to minimize the number of clashes as each of them can potentially cause a burst occassionally to...

Dynamic routing and wavelength assignment in all-optical networks is a well-known problem for which many reasonably well working heuristic algorithms have been proposed in the literature. The problem can be also approached in the Markov decision process (MDP) framework. Due to the huge size of the state space only heuristic algorithms are viable in...

Maximization of the single hop traffic has been proposed as an computationally feasible approximation to the logical topology design problem in wavelength routed networks. In this paper we consider a class of greedy heuristic algo- rithms for solving the maximization of the single hop traf- fic problem. The considered heuristics configure one light...

In this paper we present an adaptive algorithm to estimate the transient blocking probability of a communication system, described by a Markov process, during a finite time interval starting from a given state. The method uses importance sampling for variance reduction and adjusts the parameters of the twisted distribution based on earlier samples....