Johanne Cohen - Academia.edu (original) (raw)

Papers by Johanne Cohen

Lecture Notes in Computer Science, 2014

RAIRO - Operations Research, 2012

In this paper, we study the problem of makespan minimization for the multiprocessor scheduling pr... more In this paper, we study the problem of makespan minimization for the multiprocessor scheduling problem in the presence of communication delays. The communication delay between two tasks i and j depends on the distance between the two processors on which these two tasks are executed. Lahlou shows that a simple polynomial-time algorithm exists when the length of the schedule is at most two (the problem becomes N P-complete when the length of the schedule is at most three). We prove that there is no polynomial-time algorithm with a performance guarantee of less than 4/3 (unless P = N P) to minimize the makespan when the network topology is a chain or ring and the precedence graph is a bipartite graph of depth one. We also develop two polynomial-time approximation algorithms with constant ratio dedicated to cases where the processor network admits a limited or unlimited number of processors.

Lecture Notes in Computer Science, 2014

Game as a Service (GaaS) 1 has recently been deployed to cope with the limitations of user device... more Game as a Service (GaaS) 1 has recently been deployed to cope with the limitations of user devices (storage, computing power, energy). The pricing schemes –essentially flat-rate– implemented by providers are simple, but the consequences of the price levels and the offer contents on demand repartition are involved. In this paper, we introduce a model for the interactions among competing network providers offering GaaS solutions, in order to study their competition in terms of prices and selection of games to propose. We illustrate some trade-offs that network provider face, between the attractiveness of a game and its incurred costs.

Operators of networks covering large areas are confronted with demands from some of their custome... more Operators of networks covering large areas are confronted with demands from some of their customers who are virtual service providers. These providers may call for the connectivity service which fulfils the specificity of their services, for instance a multicast transition with allocated bandwidth. On the other hand, network operators want to make profit by trading the connectivity service of requested quality to their customers and to limit their infrastructure investments (or do not invest anything at all). We focus on circuit switching optical networks and work on repetitive multicast demands whose source and destinations are {\em \`a priori} known by an operator. He may therefore have corresponding trees "ready to be allocated" and adapt his network infrastructure according to these recurrent transmissions. This adjustment consists in setting available branching routers in the selected nodes of a predefined tree. The branching nodes are opto-electronic nodes which are ...

Nowadays, with the abundance of diverse air interfaces in the same operating area, advanced Radio... more Nowadays, with the abundance of diverse air interfaces in the same operating area, advanced Radio Resource Management (RRM) is vital to take advantage of the available system resources. In such a scenario, a mobile user will be able to connect concurrently to different wireless access networks. In this paper, we consider the downlink of a hybrid network with two broadband Radio Access Technologies (RAT): WiMAX and WiFi. Two approaches are proposed to load balance the traffic of every user between the two available RATs: an individual approach where mobile users selfishly strive to improve their performance and a global approach where resource allocation is made in a way to satisfy all mobile users. We devise for the individual approach a fully distributed resource management scheme portrayed as a non-cooperative game. We characterize the Nash equilibriums of the proposed RRM game and put forward a decentralized algorithm based on replicator dynamics to achieve those equilibriums. In...

Population protocols have been introduced by Angluin et al. as a model in which n passively mobil... more Population protocols have been introduced by Angluin et al. as a model in which n passively mobile anonymous finite-state agents stably compute a predicate on the multiset of their inputs via interactions by pairs. The model has been extended by Guerraoui and Ruppert to yield the community protocol models where agents have unique identifiers but may only store a finite number of the identifiers they already heard about. The population protocol models can only compute semi-linear predicates, whereas in the community protocol model the whole community of agents provides collectively the power of a Turing machine with a O(n log n) space. We consider variations on the above models and we obtain a whole landscape that covers and extends already known results: By considering the case of homonyms, that is to say the case when several agents may share the same identifier, we provide a hierarchy that goes from the case of no identifier (i.e. a single one for all, i.e. the population protocol...

Assuring Quality of Service (QoS) over multiple Network Service Providers (NSPs) requires to nego... more Assuring Quality of Service (QoS) over multiple Network Service Providers (NSPs) requires to negotiate QoS contracts as Service Level Agreements (SLAs) between NSPs. The goal of an NSP is to maximize its revenues by selling as much as possible SLAs. However, provisioning too much SLAs might increase the risk of violations of the committed QoS thus impacting the NSP's reputation. In order to determine the appropriate provisioning strategies, we propose to extend existing solutions based on Reinforcement Learning with reputation-awareness so that NSPs maximize their revenues.

In wireless access network optimization, today's main challenges reside in traffic offload an... more In wireless access network optimization, today's main challenges reside in traffic offload and in the improvement of both capacity and coverage networks. The mobile operators are interested in solving their localized coverage and capacity problems in areas where the macro network signal is not able to serve the demand for mobile data. Thus, the major issue for mobile operators is to find the best solution at reasonable expanses. The femto cell seems to be the answer to this problematic. In this work, we focus on the problem of sharing femto access between a same mobile operator's customers. A paradigm for bandwidth sharing management added to a TBAS model for exchanging connectivity is proposed for a fair sharing connectivity system ensuring QoS. This paper focuses on an economic model based on FON model and considers the sharing femto access problem as a problem divided into to 2 levels: a game restricted to service requesters customers (SRCs) and a second game restricted t...

Concurrency and Computation Practice and Experience

In this paper we consider the problem of scheduling on computing platforms composed of several in... more In this paper we consider the problem of scheduling on computing platforms composed of several independent organizations, known as the Multi-Organization Scheduling Problem (MOSP). Each organization provides both resources and jobs and follows its own objectives. We are interested in the best way to minimize the makespan on the entire platform when the organizations behave in a selfish way. We study the complexity of the MOSP problem with two different local objectives --- makespan and average completion time --- and show that MOSP is strongly NP-Hard in both cases. We formally define a selfishness notion, by means of restrictions on the schedules. We prove that selfish behavior imposes a lower bound of 2 on the approximation ratio for the global makespan. We present various approximation algorithms of ratio 2 which validate selfishness restrictions. These algorithms are experimentally evaluated through simulation, exhibiting good average performances and presenting good fairness to...

Computer Networks

Operators of networks covering large areas are confronted with demands from some of their custome... more Operators of networks covering large areas are confronted with demands from some of their customers who are virtual service providers. These providers may call for the connectivity service which fulfills the specificity of their services, for instance a multicast transmission with allocated bandwidth. On the other hand, network operators want to make profit by trading the connectivity service of requested quality to their customers and to limit their infrastructure investments (or do not invest anything at all).We focus on circuit switching optical networks and work on repetitive multicast demands whose source and destinations are à priori known by an operator. He may therefore have corresponding trees “ready to be allocated” and adapt his network infrastructure according to these recurrent transmissions. This adjustment consists in setting available branching routers in the selected nodes of a predefined tree. The branching nodes are opto-electronic nodes which are able to duplicat...

We study coordination mechanisms for Scheduling Games (with unrelated machines). In these games, ... more We study coordination mechanisms for Scheduling Games (with unrelated machines). In these games, each job represents a player, who needs to choose a machine for its execution, and intends to complete earliest possible. Our goal is to design scheduling policies that always admit a pure Nash equilibrium and guarantee a small price of anarchy for the l_k-norm social cost --- the objective balances overall quality of service and fairness. We consider policies with different amount of knowledge about jobs: non-clairvoyant, strongly-local and local. The analysis relies on the smooth argument together with adequate inequalities, called smooth inequalities. With this unified framework, we are able to prove the following results. First, we study the inefficiency in l_k-norm social costs of a strongly-local policy SPT and a non-clairvoyant policy EQUI. We show that the price of anarchy of policy SPT is O(k). We also prove a lower bound of Omega(k/log k) for all deterministic, non-preemptive, ...

Concurrency and Computation Practice and Experience

On resource sharing platforms, the execution of the jobs submitted by users is usually controlled... more On resource sharing platforms, the execution of the jobs submitted by users is usually controlled by a centralized global scheduler. It determines efficient schedules regarding some common objective function that all organizations agree with (for instance, maximizing the utilization of the entire platform). However, in practice, each organization is mostly interested in the performance obtained for its own jobs. We study the price that the collectivity must pay in order to allow independence to selfish, self-governing organizations, so they can choose the best schedules for their own jobs. In other words, we are interested in analyzing the costs on the global performance inflicted by the decentralization of scheduling policies. We present a game-theoretic model for the problem and the associated coordination mechanisms developed to reduce the cost of the decentralization of the decision-making process. The main contribution is to show (in theory and practice) how to devise pure Nash...

Owing to the proliferation of different Radio Access Technologies (RAT) in the same operating are... more Owing to the proliferation of different Radio Access Technologies (RAT) in the same operating area, a mobile user is capable of connecting concomitantly to diverse wireless networks in order to meet more easily its target performance. In this paper, we consider the downlink of a multi-class hybrid network with two RATs: WiMAX and 3G LTE. We put forward a semi-distributed Radio Resource Management (RRM) scheme for elastic traffic where both the system and mobile users intervene in the resource management policy. The proposed scheduling scheme is original in the sense that users with elastic traffic have a counterintuitive behavior: they will try to occupy the least amount possible of bandwidth to accommodate QoS stringent streaming traffic. A non-cooperative game is used to load balance the traffic of elastic users between the two available RATs aiming at minimizing their bandwidth consumption. We characterize the Nash Equilibriums (NE) of the RRM game and study the efficiency of a b...

We conduct a game theoretic analysis on the problem of scheduling jobs on computing platforms com... more We conduct a game theoretic analysis on the problem of scheduling jobs on computing platforms composed of several independent and selfish organizations, known as the Multi-Organization Scheduling Problem (MOSP). Each organization shares resources and jobs with others, expecting to decrease the makespan of its own jobs. We modeled MOSP as a non-cooperative game where each agent is responsible for assigning all jobs belonging to a particular organization to the available processors. The local scheduling of these jobs is defined by coordination mechanisms that first prioritize local jobs and then schedule the jobs from others according to some given priority. When different priorities are given individually to the jobs — like in classical scheduling algorithms such as LPT or SPT — then no pure e-approximate equilibrium is possible for values of e less than 2. We also prove that even deciding whether a given instance admits or not a pure Nash equilibrium is co-NP hard. When these priori...

International Journal of High Performance Computing and Networking, 2006

We deal with a way of realizing real-time communications required by a distributed interactive si... more We deal with a way of realizing real-time communications required by a distributed interactive simulation (DIS), using multipoint communication technics. These technics would be the basic principles of the data distributed management (DDM) of the simulation tool. We focus here on a classical interactive game between participants distributed on a geographic map, where each participant is associated to a square cell on it. The needs of communication between participants (i.e., if the associated cells overlap) are represented by a graph called neighborhood graph. The problem we deal with consists in covering efficiently the neighborhood graphs by groups of nodes (such that each edge is covered by at least one group), and in allocating in the target network a subtree with a given bandwidth to each group. After giving the formal definition of the considered problem, we show that it is NP-complete. Then, we give some lower bounds technics. Finally, we give two heuristics to solve this problem and we analyse them. ). Her main research interest include the study of telecommunication networks and approximation methods for hard combinatorial problems.

Proceedings of the 4th International ICST Conference on Performance Evaluation Methodologies and Tools, 2009

We study how to reach a Nash equilibrium in a load balancing scenario where each task is managed ... more We study how to reach a Nash equilibrium in a load balancing scenario where each task is managed by a selfish agent and attempts to migrate to a machine which will minimize its cost. The cost of a machine is a function of the load on it. The load on a machine is the sum of the weights of the jobs running on it. We prove that Nash equilibria can be learned on that games with incomplete information, using some Lyapunov techniques.

Lecture Notes in Computer Science, 2008

We consider the problem of learning equilibria in a well known game theoretic traffic model due t... more We consider the problem of learning equilibria in a well known game theoretic traffic model due to Wardrop. We consider a distributed learning algorithm that we prove to converge to equilibria. The proof of convergence is based on a differential equation governing the global macroscopic evolution of the system, inferred from the local microscopic evolutions of agents. We prove that the differential equation converges with the help of Lyapunov techniques.

Lecture Notes in Computer Science, 2014

RAIRO - Operations Research, 2012

In this paper, we study the problem of makespan minimization for the multiprocessor scheduling pr... more In this paper, we study the problem of makespan minimization for the multiprocessor scheduling problem in the presence of communication delays. The communication delay between two tasks i and j depends on the distance between the two processors on which these two tasks are executed. Lahlou shows that a simple polynomial-time algorithm exists when the length of the schedule is at most two (the problem becomes N P-complete when the length of the schedule is at most three). We prove that there is no polynomial-time algorithm with a performance guarantee of less than 4/3 (unless P = N P) to minimize the makespan when the network topology is a chain or ring and the precedence graph is a bipartite graph of depth one. We also develop two polynomial-time approximation algorithms with constant ratio dedicated to cases where the processor network admits a limited or unlimited number of processors.

Lecture Notes in Computer Science, 2014

Game as a Service (GaaS) 1 has recently been deployed to cope with the limitations of user device... more Game as a Service (GaaS) 1 has recently been deployed to cope with the limitations of user devices (storage, computing power, energy). The pricing schemes –essentially flat-rate– implemented by providers are simple, but the consequences of the price levels and the offer contents on demand repartition are involved. In this paper, we introduce a model for the interactions among competing network providers offering GaaS solutions, in order to study their competition in terms of prices and selection of games to propose. We illustrate some trade-offs that network provider face, between the attractiveness of a game and its incurred costs.

Operators of networks covering large areas are confronted with demands from some of their custome... more Operators of networks covering large areas are confronted with demands from some of their customers who are virtual service providers. These providers may call for the connectivity service which fulfils the specificity of their services, for instance a multicast transition with allocated bandwidth. On the other hand, network operators want to make profit by trading the connectivity service of requested quality to their customers and to limit their infrastructure investments (or do not invest anything at all). We focus on circuit switching optical networks and work on repetitive multicast demands whose source and destinations are {\em \`a priori} known by an operator. He may therefore have corresponding trees "ready to be allocated" and adapt his network infrastructure according to these recurrent transmissions. This adjustment consists in setting available branching routers in the selected nodes of a predefined tree. The branching nodes are opto-electronic nodes which are ...

Nowadays, with the abundance of diverse air interfaces in the same operating area, advanced Radio... more Nowadays, with the abundance of diverse air interfaces in the same operating area, advanced Radio Resource Management (RRM) is vital to take advantage of the available system resources. In such a scenario, a mobile user will be able to connect concurrently to different wireless access networks. In this paper, we consider the downlink of a hybrid network with two broadband Radio Access Technologies (RAT): WiMAX and WiFi. Two approaches are proposed to load balance the traffic of every user between the two available RATs: an individual approach where mobile users selfishly strive to improve their performance and a global approach where resource allocation is made in a way to satisfy all mobile users. We devise for the individual approach a fully distributed resource management scheme portrayed as a non-cooperative game. We characterize the Nash equilibriums of the proposed RRM game and put forward a decentralized algorithm based on replicator dynamics to achieve those equilibriums. In...

Population protocols have been introduced by Angluin et al. as a model in which n passively mobil... more Population protocols have been introduced by Angluin et al. as a model in which n passively mobile anonymous finite-state agents stably compute a predicate on the multiset of their inputs via interactions by pairs. The model has been extended by Guerraoui and Ruppert to yield the community protocol models where agents have unique identifiers but may only store a finite number of the identifiers they already heard about. The population protocol models can only compute semi-linear predicates, whereas in the community protocol model the whole community of agents provides collectively the power of a Turing machine with a O(n log n) space. We consider variations on the above models and we obtain a whole landscape that covers and extends already known results: By considering the case of homonyms, that is to say the case when several agents may share the same identifier, we provide a hierarchy that goes from the case of no identifier (i.e. a single one for all, i.e. the population protocol...

Assuring Quality of Service (QoS) over multiple Network Service Providers (NSPs) requires to nego... more Assuring Quality of Service (QoS) over multiple Network Service Providers (NSPs) requires to negotiate QoS contracts as Service Level Agreements (SLAs) between NSPs. The goal of an NSP is to maximize its revenues by selling as much as possible SLAs. However, provisioning too much SLAs might increase the risk of violations of the committed QoS thus impacting the NSP's reputation. In order to determine the appropriate provisioning strategies, we propose to extend existing solutions based on Reinforcement Learning with reputation-awareness so that NSPs maximize their revenues.

In wireless access network optimization, today's main challenges reside in traffic offload an... more In wireless access network optimization, today's main challenges reside in traffic offload and in the improvement of both capacity and coverage networks. The mobile operators are interested in solving their localized coverage and capacity problems in areas where the macro network signal is not able to serve the demand for mobile data. Thus, the major issue for mobile operators is to find the best solution at reasonable expanses. The femto cell seems to be the answer to this problematic. In this work, we focus on the problem of sharing femto access between a same mobile operator's customers. A paradigm for bandwidth sharing management added to a TBAS model for exchanging connectivity is proposed for a fair sharing connectivity system ensuring QoS. This paper focuses on an economic model based on FON model and considers the sharing femto access problem as a problem divided into to 2 levels: a game restricted to service requesters customers (SRCs) and a second game restricted t...

Concurrency and Computation Practice and Experience

In this paper we consider the problem of scheduling on computing platforms composed of several in... more In this paper we consider the problem of scheduling on computing platforms composed of several independent organizations, known as the Multi-Organization Scheduling Problem (MOSP). Each organization provides both resources and jobs and follows its own objectives. We are interested in the best way to minimize the makespan on the entire platform when the organizations behave in a selfish way. We study the complexity of the MOSP problem with two different local objectives --- makespan and average completion time --- and show that MOSP is strongly NP-Hard in both cases. We formally define a selfishness notion, by means of restrictions on the schedules. We prove that selfish behavior imposes a lower bound of 2 on the approximation ratio for the global makespan. We present various approximation algorithms of ratio 2 which validate selfishness restrictions. These algorithms are experimentally evaluated through simulation, exhibiting good average performances and presenting good fairness to...

Computer Networks

Operators of networks covering large areas are confronted with demands from some of their custome... more Operators of networks covering large areas are confronted with demands from some of their customers who are virtual service providers. These providers may call for the connectivity service which fulfills the specificity of their services, for instance a multicast transmission with allocated bandwidth. On the other hand, network operators want to make profit by trading the connectivity service of requested quality to their customers and to limit their infrastructure investments (or do not invest anything at all).We focus on circuit switching optical networks and work on repetitive multicast demands whose source and destinations are à priori known by an operator. He may therefore have corresponding trees “ready to be allocated” and adapt his network infrastructure according to these recurrent transmissions. This adjustment consists in setting available branching routers in the selected nodes of a predefined tree. The branching nodes are opto-electronic nodes which are able to duplicat...

We study coordination mechanisms for Scheduling Games (with unrelated machines). In these games, ... more We study coordination mechanisms for Scheduling Games (with unrelated machines). In these games, each job represents a player, who needs to choose a machine for its execution, and intends to complete earliest possible. Our goal is to design scheduling policies that always admit a pure Nash equilibrium and guarantee a small price of anarchy for the l_k-norm social cost --- the objective balances overall quality of service and fairness. We consider policies with different amount of knowledge about jobs: non-clairvoyant, strongly-local and local. The analysis relies on the smooth argument together with adequate inequalities, called smooth inequalities. With this unified framework, we are able to prove the following results. First, we study the inefficiency in l_k-norm social costs of a strongly-local policy SPT and a non-clairvoyant policy EQUI. We show that the price of anarchy of policy SPT is O(k). We also prove a lower bound of Omega(k/log k) for all deterministic, non-preemptive, ...

Concurrency and Computation Practice and Experience

On resource sharing platforms, the execution of the jobs submitted by users is usually controlled... more On resource sharing platforms, the execution of the jobs submitted by users is usually controlled by a centralized global scheduler. It determines efficient schedules regarding some common objective function that all organizations agree with (for instance, maximizing the utilization of the entire platform). However, in practice, each organization is mostly interested in the performance obtained for its own jobs. We study the price that the collectivity must pay in order to allow independence to selfish, self-governing organizations, so they can choose the best schedules for their own jobs. In other words, we are interested in analyzing the costs on the global performance inflicted by the decentralization of scheduling policies. We present a game-theoretic model for the problem and the associated coordination mechanisms developed to reduce the cost of the decentralization of the decision-making process. The main contribution is to show (in theory and practice) how to devise pure Nash...

Owing to the proliferation of different Radio Access Technologies (RAT) in the same operating are... more Owing to the proliferation of different Radio Access Technologies (RAT) in the same operating area, a mobile user is capable of connecting concomitantly to diverse wireless networks in order to meet more easily its target performance. In this paper, we consider the downlink of a multi-class hybrid network with two RATs: WiMAX and 3G LTE. We put forward a semi-distributed Radio Resource Management (RRM) scheme for elastic traffic where both the system and mobile users intervene in the resource management policy. The proposed scheduling scheme is original in the sense that users with elastic traffic have a counterintuitive behavior: they will try to occupy the least amount possible of bandwidth to accommodate QoS stringent streaming traffic. A non-cooperative game is used to load balance the traffic of elastic users between the two available RATs aiming at minimizing their bandwidth consumption. We characterize the Nash Equilibriums (NE) of the RRM game and study the efficiency of a b...

We conduct a game theoretic analysis on the problem of scheduling jobs on computing platforms com... more We conduct a game theoretic analysis on the problem of scheduling jobs on computing platforms composed of several independent and selfish organizations, known as the Multi-Organization Scheduling Problem (MOSP). Each organization shares resources and jobs with others, expecting to decrease the makespan of its own jobs. We modeled MOSP as a non-cooperative game where each agent is responsible for assigning all jobs belonging to a particular organization to the available processors. The local scheduling of these jobs is defined by coordination mechanisms that first prioritize local jobs and then schedule the jobs from others according to some given priority. When different priorities are given individually to the jobs — like in classical scheduling algorithms such as LPT or SPT — then no pure e-approximate equilibrium is possible for values of e less than 2. We also prove that even deciding whether a given instance admits or not a pure Nash equilibrium is co-NP hard. When these priori...

International Journal of High Performance Computing and Networking, 2006

We deal with a way of realizing real-time communications required by a distributed interactive si... more We deal with a way of realizing real-time communications required by a distributed interactive simulation (DIS), using multipoint communication technics. These technics would be the basic principles of the data distributed management (DDM) of the simulation tool. We focus here on a classical interactive game between participants distributed on a geographic map, where each participant is associated to a square cell on it. The needs of communication between participants (i.e., if the associated cells overlap) are represented by a graph called neighborhood graph. The problem we deal with consists in covering efficiently the neighborhood graphs by groups of nodes (such that each edge is covered by at least one group), and in allocating in the target network a subtree with a given bandwidth to each group. After giving the formal definition of the considered problem, we show that it is NP-complete. Then, we give some lower bounds technics. Finally, we give two heuristics to solve this problem and we analyse them. ). Her main research interest include the study of telecommunication networks and approximation methods for hard combinatorial problems.

Proceedings of the 4th International ICST Conference on Performance Evaluation Methodologies and Tools, 2009

We study how to reach a Nash equilibrium in a load balancing scenario where each task is managed ... more We study how to reach a Nash equilibrium in a load balancing scenario where each task is managed by a selfish agent and attempts to migrate to a machine which will minimize its cost. The cost of a machine is a function of the load on it. The load on a machine is the sum of the weights of the jobs running on it. We prove that Nash equilibria can be learned on that games with incomplete information, using some Lyapunov techniques.

Lecture Notes in Computer Science, 2008

We consider the problem of learning equilibria in a well known game theoretic traffic model due t... more We consider the problem of learning equilibria in a well known game theoretic traffic model due to Wardrop. We consider a distributed learning algorithm that we prove to converge to equilibria. The proof of convergence is based on a differential equation governing the global macroscopic evolution of the system, inferred from the local microscopic evolutions of agents. We prove that the differential equation converges with the help of Lyapunov techniques.