Federations of Infrastructure Service Providers (InfSPs) have emerged as an effective practice to cater for demanding applications in cloud- and edge-computing environments, by overcoming the limitations of each such provider (in terms of available resources, geographic coverage etc.). In this paper, we develop innovative mechanisms and policies for the decentralized management of InfSP federations, considering the needs both for resource allocation meeting service composition requirements, and for performing a fair and efficient distribution of profits. Resource allocation is performed by means of a greedy algorithm that includes an opportunistic and an adjustment phase. To determine the payments of participating InfSPs, we introduce an innovative sealed-bid reverse auction that combines VCG and first-price auction mechanisms. This hybrid mechanism, coupled with a sophisticated federation wallet management mechanism, serves as a revenue sink for the federation. Notably, our approach exhibits several nice properties, including efficiency, long-term budget balance, truthfulness and individual rationality. Furthermore, we present how our solution can be implemented in a decentralized, privacy-preserving and trustworthy manner over a blockchain, achieving bid privacy, non-repudiation and public verifiability. Our approach is evaluated by means of extensive experiments, the results of which reveal that resource allocation is either optimal or very close thereto, while individual InfSP profits as well as total federation profits are maximized. The results underscore the inadequacy of relying solely on the VCG mechanism for budget balance, justifying the necessity of our sophisticated hybrid solution.

Network slicing allows Mobile Network Operators (MNOs) to partition their physical infrastructure into multiple virtual logical networks, enabling the simultaneous servicing of applications with diverse Quality of Service characteristics. In this paper, we introduce and evaluate economic models and policies for the provisioning, across multiple MNOs, of network slice services to Application Providers. We introduce a Network-Slice-as-a-Service model that maps the service offered by a network slice to requirements on virtualized resources. The placement of virtualized resources over the physical infrastructure of MNOs is determined by an embedding problem, formulated as a Mixed Integer Program. We investigate the embedding under: (i) centralized approaches, where a central Broker determines the embedding for all network slice requests, and (ii) peer-to-peer approaches, where each MNO determines the embedding for the sub-set of network slice requests coming from its own customers. We introduce policies for cooperative modes, with the objective of total profit maximization, and for “coopetitive” (cooperative competition) modes, where MNOs aim to maximize their individual profits. The numerical results reveal that MNOs can maximize their aggregate and individual profits under any approach or mode, if they comply with the proposed policies.

Sharing economy is a game-changing business paradigm that is currently permeating several industrial sectors. This paper aims to build a fundamental theory of the sharing economy of the computational capacity resource of Cloud Service Providers (CSPs). CSPs aim to cost-efficient serve geographically dispersed customers that often request computational resource-demanding services. The formation of CSP federations arises as an effective means to manage these diverse and time-varying service requests. In this paper, we introduce innovative federation models and policies for profitable federations that also achieve adequate QoS for their customers. Taking in account the flexible cloud computing service model, we abstract the virtualized infrastructure of each CSP to an M/M/1 queueing system, we formulate the CSP revenue and cost functions, and we study the task forwarding-based (TF) and the capacity sharing-based (CS) federation approaches. Under TF, each CSP may forward part of its workload to other federated CSPs, while under CS each CSP may share parts of its computational infrastructure with others. For both approaches, we propose two operation modes with different degree of CSPs' cooperation: (i) the joint business mode, where the CSPs fully cooperate: they jointly decide on the federation policies that maximize the total federation profit which is shared fairly among them; (ii) the reward-driven mode, where selfinterested CSPs participate in a game: they adjust their responses to federation policies aiming to maximize their individual profits. The results reveal that our policies lead to effective federations, which are beneficial both for CSPs and for customers.

This paper contains an original classification of the 5G wholesale services and a novel proposal for the respective pricing schemes for these services. Motivated by the envisioned 5G customer services, this paper introduces the services that need to be offered, orchestrated and traded among infrastructure and service providers, so that end-to-end 5G services can be efficiently orchestrated and provisioned. The paper proposes and assesses candidate pricing schemes for these services.

In this paper we introduce the Inter-Cloud Communication Traffic Management mechanism (ICC). ICC performs rate control over the ISP transit link(s) aiming to attain a target reduction of transit cost. ICC reduces ISPs transit charge by shaping a portion of the inter-domain traffic (e.g. delay-tolerant inter-cloud traffic), which has been marked as time-shiftable by the traffic source, i.e. by the business customer of the ISP such as a cloud/data center. In particular, ICC reduces the transmission rate of marked traffic at peak 5-min billable intervals and increases it at off-peak, acting at even shorter timescales according to a novel rate-adaptation algorithm. We evaluate numerically ICC by employing real traffic traces. The results reveal that ICC can significantly reduce the ISP transit charge and thus can be a promising and practically applicable solution for ISPs while also being beneficial for their customers whom the ISP should incentivize by sharing its savings.

The Internet has evolved towards a unique technology base for creating value-added services with worldwide connectivity. These services, enabled by the increasing bandwidth of access networks, result in new high-performance applications (e.g. e-health, high definition video streaming, network gaming etc.) To support and materialize the value of these emerging services, it is important for the operators to be able to provide some form of Quality of Service (QoS) assurance. This paper presents the main economic issues regarding the efficient provisioning of such services in inter-domain level and overviews some candidate economic mechanisms and game-theoretic tools that could be adopted.

We study flow optimization in cognitive Wireless Mesh Networks (WMN) that lease bandwidth from a collocated cellular network (CN). We develop a distributed mechanism that such WMNs can use to optimize their traffic given their own utilities and costs and the costs that the CN is charging for the use of its channels. We evaluate the mechanism in a synthetic scenario using the topology of a deployed WMN and time-varying traffic requirements induced from real-life vehicle mobility traces. Simulations show that the mechanism can track well the changes in the problem parameters, continuously keeping close to the optimum operation point.

Abstract The concept of service‐oriented architectures (SOA) has recently emerged as a design principle for the next generation of IT solutions. The main idea behind SOA lies in treating software applications as composed of simpler, cooperating services, implemented by components that communicate over networks through open standards. Such a modular structure is expected to question the effectiveness of current business models for building and distributing software, which generate revenue through license fees. Under SOA, the license fee is replaced by more flexible schemes, whereby users are charged according to their actual usage of services and the corresponding hardware infrastructure. Our work constitutes a first step toward exploring the economic aspects of a market for software components, as well as the incentives of software manufacturers to support this trend. In particular, we examine the factors that affect profitability in an open market for services, we build simple models to predict and explain market growth and we also suggest ways to accelerate this growth, while also achieving a higher level of economic efficiency. Copyright © 2009 John Wiley & Sons, Ltd.

Hidden information is a critical issue for the successful delivery of SLAs in grid systems. It arises when the agents (hardware and software resources) employed to serve a task belong to multiple administrative domains, thus rendering monitoring of remote resource provision absent or unreliable. Therefore, the grid service broker can often observe only the outcome of the collective effort of groups of agents rather than their individual efforts, which makes it hard to identify cases of free-riding or low- performing agents. In this paper, we first identify cases of hidden information in grid systems and explain why they cannot be handled satisfactorily by the existing accounting systems. Second, we develop and evaluate a reputation-based mechanism enabling the grid service broker to deal effectively with hidden information. Our mechanism maintains a reputation metric for each agent; we propose and evaluate several approaches on how to update this metric based only on the observations of collective outcomes.

The effectiveness of online feedback mechanisms for rating the performance of providers in electronic markets is vulnerable to the submission of dishonest ratings. In this paper, we deal with how to elicit honest such ratings in a competitive electronic market where each participant can occasionally act both as provider and as client. We assume that each service provision is rated by both parties involved; only upon agreement, this rating is included in the calculation of reputation for the provider's performance. We first study as a single-shot game the effectiveness of inducing, upon evidence of lying (i.e. disagreement of the submitted feedback), fixed fines to both transacted parties, yet different ones for the provider and the client. We prove that the submission of honest feedback can be a stable equilibrium for the whole market under certain initial system conditions. Then, we refine our game-model for repeated transactions and calculate proper different reputation-based fines for lying. These fines enable the submission of honest feedback as a stable Nash equilibrium of the repeated game and reduce the social losses due to unfair punishments. Finally, we argue that our model is appropriate for analyzing actual electronic markets, and we investigate the impact of employing our approach to the feedback schemes of such markets.

We propose a mechanism for providing the incentives for reporting truthful feedback in a peer-to-peer system for exchanging services. This mechanism is to complement reputation mechanisms that employ ratings' feedback on the various transactions in order to provide incentives to peers for offering better services to others. Under our approach, both transacting peers (rather than just the client) submit ratings on performance of their mutual transaction. If these are in disagreement, then both transacting peers are punished, since such an occasion is a sign that one of them is tying. The severity of each peer's punishment is determined by his corresponding non-credibility metric; this is maintained by the mechanism and evolves according to the peer's record. When under punishment, a peer is not allowed to transact with others. We present the results of a multitude of experiments of dynamically evolving peer-to-peer systems. The results show clearly that our mechanism detects and isolates effectively liar peers, while rendering lying costly. Also, our mechanism diminishes the efficiency losses induced to sincere peers by the presence of large subsets of the population of peers that provide their ratings either falsely or according to various unfair strategies. Finally, we explain how our approach can be implemented in practical cases of peer-to-peer systems.

We revisit the network-wide Progressive Second Price auction (PSP) proposed by Lazar and Semret in cite2. In this mechanism, each bidder submits the same bid in each link of the path he is interested in, taking into account the overall competition in the path. We propose a new strategy in which each bidder apportions his total bid-price in the various links, while taking into account the competition in each link separately. We have carried out a wide variety of experiments and compare our strategy with the original one with respect to efficiency and bidders' net benefit. We show that our strategy yields an outcome closer to the optimal social welfare as well as higher expected net benefit for bidders in most of the cases.

Peer-to-peer environments have become popular as a framework for exchange of services. In these environments, certain peers may fail to provide. their services.. Reputation can be a proper means of discovering low-performing peers, without affecting significantly inherent characteristics of peer-to-peer environments, such as anonymity and privacy. However, the accurate calculation of the reputation metrics may not be sufficient to provide the right incentives to peers. In this paper, we show that the straightforward approach for peers to exploit the reputation metrics (i.e. by just selecting as a providing peer the one with the highest reputation) may lead to unexpectedly low efficiency for high-performing peers. We argue and justify experimentally that the calculation of the reputation values has to be complemented by reputation-based policies that define the pairs of peers eligible to interact. We introduce two orthogonal dimensions constituting the reputation based policies: provider selection and contention resolution. We argue and show by means of simulation experiments that both these dimensions have a significant impact to the achieved efficiency of the peers. We also investigate experimentally the achievable efficiency of specific reputation-based policies for the case of short-lived peers of two different fixed-strategy types. Finally, we deal with the efficient computation of the reputation value by means of aggregation of the ratings feedback provided by the peers. We propose that this can be accomplished by aggregating only a small randomly selected subset of this feedback. Simulation experiments indicate that this approach indeed leads to the fast and accurate calculation of the reputation values even if the peer-to-peer population is renewed with a high rate.

For a multiplexer fed by a large number of sources, we derive conditions under which a given subset of the sources can be substituted for a single source while preserving the buffer overflow probability and the dominant timescales of buffer overflows. This notion of traffic equivalence is stronger than simple effective bandwidth equality and depends on the multiplexing context. We propose several applications of the above traffic substitution conditions. First, we show that fractional Brownian motion as a single source substitute can effectively model a large number of multiplexed sources using information obtained purely from traffic traces; this has direct application to simple but accurate traffic generation. Second, we focus on dynamic (i.e., on-line) estimation of available capacity and buffer overflow probability. This requires the solution of a double optimization problem expressed in terms of functions whose values are obtained from time averages of the traffic traces over a large range of timescales. We show how to solve this problem on-line by reducing it to the calculation of a fixed-point equation that can be solved iteratively by combining traffic substitution using fractional Brownian motion with dynamic measurements of the actual traffic. We have validated this approach by extensive experimentation with large numbers of real traffic sources that are fed to a high bandwidth link, and comparing our on-line estimation of available capacity and the resulting dynamic call admission control with other existing approaches. The superior accuracy of our approach also suggests that taking the buffer size into account, as does our on-line algorithm, may be vital for achieving approximations of practical interest.

Energy dissipated in on-chip caches represents a substantial portion in the energy budget of today's processors. Extrapolating current trends, this portion is likely to increase in the near future, since the devices devoted to the caches occupy an increasingly larger percentage of the total area of the chip. We extend the work proposed by J. Kin et al. (1997), in which an extra, small cache (called filter cache) is inserted between the CPU data path and the L1 cache and serves to filter most of the references initiated from the CPU. In our scheme, the compiler is used to generate code that exploits the new memory hierarchy and reduces the possibility of a miss in the extra cache. Experimental results across a wide range of SPEC95 benchmarks show that this cache, which we call L-Cache, has a small performance overhead with respect to the scheme without any extra caches, and provides substantial energy savings. The L-Cache is placed between the CPU and the I-Cache. The D-Cache subsystem is not modified. Since the L-Cache is much smaller, and thus, has a smaller access time than the I-Cache, this scheme can also be used for performance improvements provided that the hit rate in the L-Cache is very high. In our experimental results, we show that the L-Cache does indeed improve performance in some cases.

For a multiplexer fed with a large number of sources, we derive conditions under which a single source can be substituted for a given subset of the sources while preserving the buffer overflow probability and the dominant time scales of buffer overflows. This equivalence is stronger than simple effective bandwidth equality and takes into account the context in which multiplexing takes place. This allows a substitution to be made for arbitrarily large proportions of the traffic without changing the operating point of the multiplexer as experienced by the rest of the traffic. It corresponds to defining a single source which is equivalent in a sense "local" to a given context, rather than equivalent in a sense which is "universal" to all contexts. The proposed methodology does not rely on traffic models and obtains the necessary information from the actual traffic traces. We study the case of fractional Brownian motion as a single source substitute and provide theoretical and experimental results that validate our approach.

In this paper we propose a technique that uses an additional mini cache located between the I-Cache and the CPU core, and buffers instructions that are nested within loops and are continuously otherwise fetched from the I-Cache. This mechanism is combined with code modifications, through the compiler, that greatly simplify the required hardware, eliminate unnecessary instruction fetching, and consequently reduce signal switching activity and the dissipated energy.

Accurate yet simple methods for traffic engineering are important for efficient dimensioning of broadband networks. The goal of this paper is to apply and evaluate large deviation techniques for traffic engineering. In particular, we employ the recently developed theory of effective bandwidths, where the effective bandwidth depends not only on the statistical characteristics of the traffic stream, but also on a link's operating point through two parameters, the space and time parameters, which are computed using the many sources asymptotic. We show that this effective bandwidth definition can accurately quantify resource usage. Furthermore, we estimate and interpret values of the space and time parameters for various mixes of real traffic demonstrating how these values can be used to clarify the effects on the link performance of the time scales of burstiness of the traffic input, of the link parameters (capacity and buffer), and of traffic control mechanisms, such as traffic shaping. Our approach relies on off-line analysis of traffic traces, the granularity of which is determined by the time parameter of the link, and our experiments involve a large set of MPEG-1 compressed video and Internet Wide Area Network (WAN) traces, as well as modeled voice traffic.

We study the communication complexity of asynchronous distributed algorithms. Such algorithms can generate excessively many messages in the worst case. Nevertheless, we show that, under certain probabilistic assumptions, the expected number of messages generated per time unit is bounded by a polynomial function of the number of processors under a very general model of distributed computation. Furthermore, for constant-degree processor graphs, the expected number of generated messages is only O(nT) , where n is the number of processors and T is the running time. We conclude that (under our model) any asynchronous algorithm with good time complexity will also have good communication complexity, on the average.

We analyze the following problem. Each node of the d-dimensional hypercube independently generates packets according to a Poisson process with rate /spl lambda/. Each of the packets is to be sent to a randomly chosen destination; each of the nodes at Hamming distance k from a packet's origin is assigned an a priori probability p/sup k/(1-p)/sup d-k/. Packets are routed under a simple greedy scheme: each of them is forced to cross the hypercube dimensions required in increasing index-order, with possible queueing at the hypercube nodes. Assuming unit packet length and no other communications taking place, we show that this scheme is stable (in steady-state) if /spl rho/>

The authors analyze the problem in which each node of the binary hypercube independently generates packets according to a Poisson process with rate lambda ; each of the packets is to be broadcast to all other nodes. Assuming unit packet length and no other communications taking place, it is observed that the system can be stable in steady-state only if the load factor rho identical to lambda (2/sup d/-1)/d satisfies rho>

We analyze the following problem: Each node of the d-

A proof is provided that a logarithmic redundancy factor is necessary for the reliable computation of the parity function by means of a network with noisy gates. This result was first stated by R.L. Dobrushin and S.I. Ortyukov (1977). However, the authors believe that the analysis given by Dobrushin and Ortyukov is not entirely correct. The authors establish the result by following the same steps and by replacing the questionable part of their analysis with entirely new arguments.>