MiddleWEdge

2nd International Workshop on Middleware for the Edge

co-located with ACM Middleware 2023 in Bologna, Italy

The growing trend of cloud computing adoption is being driven by the pursuit of reducing upfront costs with hardware and staff, which is further accentuated with the on-the-fly adjustment of the computational needs through its natural resource elasticity. The increasing pervasiveness of cloud computing is being supported by just a handful of vendors, such as Amazon, Alibaba, Baidu, Google and Microsoft, which creates the perfect conditions for cartelization of these services.

At the same time, ongoing efforts are trying to leverage these facilities with edge computing, where computation is performed, at least partially, nearer to the data sources, sensors and user's devices, in a concerted effort to reduce latency and the amount of data transfer to the cloud. The use of these solutions is driven by the rising popularity of IoT devices and cyber physical systems, where a large number of sensors and actuators are contributing to the computing continuum and highlighting challenges related to the robustness, and independence of edge sites.

PROGRAM (December 11th):

·       09h00 – Keynote: Challenges and Opportunities in Edge-Cloud Application Development through QoS-aware Middleware Solutions. Armir Bujari (University of Bologna)

·       10h00 – Offloading Real-Time Tasks in IIoT Environments under Consideration of Networking Uncertainties. Ilja Behnke, Philipp Wiesner, Paul Voelker, Odej Kao (Technische Universität Berlin)

·       10h30 – 11h00 – Coffee Break

·       11h00 – On the adversarial robustness of full integer quantized TinyML models at the edge. Davy Preuveneers, Willem Verheyen, Sander Joos, Wouter Joosen (KU Leuven)

·       11h30 – Performance Characterization of MQTT Brokers in a Device-Local Edge Deployment. Guillaume Simard, Cédric Melançon, Patrick Cardinal, Julien Gascon-Samson (ÉTS Montréal)

·       12h00 – Enoki: Stateful Distributed FaaS from Edge to Cloud. Tobias Pfandzelter, David Bermbach (Technische Universität Berlin & Einstein Center Digital Future)

·       12h30 – Closing

CALL FOR PAPERS:

The increasing adoption of both cloud and edge computing is being driven by classical architectures which are based on well-known centralized models to manage the life cycle of hard and software resources. The introduction of geo-distributed or decentralized clouds can offer an alternative to established commercial providers, allowing for a democratization of access to computation while acting as a reinforcement to self-sovereign, particularly in the EU. Moreover, as shown in recent large scale events, such as the COVID-19 outburst, centralized infrastructures pose a significant threat to the sustainability of our societal fabric by targeting our economy, security and personal freedoms.

The research community and business can jumpstart the edge computing paradigm and provide practical solutions for the market by offering novel models and mechanisms that ensure secure, effective, and safe management of compute, data, and communications.

The increasing adoption of both cloud and edge computing is being driven by classical architectures which are based on well-known centralized models to manage the life cycle of hard and software resources. The introduction of geo-distributed or decentralized clouds can offer an alternative to established commercial providers, allowing for a democratization of access to computation while acting as a reinforcement to self-sovereign, namely in the EU. Moreover, as shown in recent large-scale events, such as the current COVID-19 outburst, centralized infrastructures pose a significant threat to the sustainability of our societal fabric by targeting our economy, security, and personal freedoms.

To overcome these limitations, the MiddleWEdge workshop aims to address the decentralization and democratization of computation and storage. By offering novel models and mechanisms to ensure secure, efficient and safe management of compute, data and communications, the research community and industry can bootstrap the edge computing paradigm and offer concrete solutions to the market. There are multiple challenges that are considered to be core to MiddleWEdge, namely:

- Identity and Access Management: Explore decentralized management, authentication and authorization as a way to remove centralized solutions, as it would become a single-point of failure forcing all nodes to trust this specific third party. Using concepts introduced by self-sovereignty solutions, such as Sovrin, to provide decentralized identifiers, e.g., persons or aliases, to provide privacy to the end users. Moreover, there is a need for novel solutions that support decentralized authorization and access control. The problem of many of current centralized and even decentralized solutions is that they store metadata and ACLs with the servers or the network overlay, and thus forcing delegation;

- Management of the resource life cycle: Similarly to the IAM challenges, the management of the resources should be revisited to decentralized approaches. More precisely, we expect that edge resources will be managed by multiple independent entities, each one in charge of controlling one part of the edge infrastructures, delivering this way independence, as well as autonomy of the resources. It is noteworthy that the management of the life cycle relies on various mechanisms including monitoring of resources, controller loops, deployment/reconfiguration techniques that should be explored;

- Scheduling: Exploring novel paths to make scheduling and service placement fully decentralized, able to seamlessly leverage devices with heterogeneous architectures and capabilities, in an efficient and scalable manner. This entails employing self-organizing algorithms continuously running, matching requests and resources preferably locally (while aiming to converge towards global balancing).; The use of reinforcement techniques can be relevant to predict future usages of the infrastructures, as well as the evolution of the needs of applications, providing important hints for the scheduling decisions, with the goal of optimising different objectives (efficiency, energy, etc.);

Topics:

- Decentralized identity management and access control

- Decentralized middleware/resource management systems

- Decentralized scheduling strategies

- Novel storage systems for edge clouds, with special focus on geo-aware storage engines

- Mechanisms for the integration of edge clouds and public clouds

- Security mechanisms for edge clouds including, including but not limited to, storage and computation

- IoT-Edge computing continuum

- Services for thin edge integration

- Privacy preserving algorithms that leverage edge clouds

- Trust and reputation for edge clouds

- Distributed ledgers for supporting edge clouds

- Lightweight replication and fault-tolerance algorithms

- Lightweight computation sandboxing for edge clouds

- Tools for testing and benchmarking for edge clouds

- Experimental deployments and applications

- Novel theoretical approaches for churn tolerance

- Networking coding approaches for edge clouds

- Programming languages for edge clouds

- P2P overlays for edge clouds

- Gossip based protocols for edge clouds

- Computational frameworks for edge clouds

- Programming models and abstractions to manage inter-cloud interactions

- Distributed coordination and cooperation for edge clouds

Organizing Committee:

• Daniel Balouek-Thomert, Inria (daniel.balouek-thomert@inria.fr)
• Felix Freitag, Polytechnic University of Catalonia (felix.freitag@upc.edu)
• Adrien Lebre, IMT Atlantique (adrien.lebre@imt-atlantique.fr)
• Luís Veiga, INESC-ID, Técnico Lisboa, ULisboa (luis.veiga@inesc-id.pt)

Program Committee:

• Paolo Bellavista (UNIBO)
• David Bermbach (TU Berlin)
• Alysson Bessani (University of Lisbon)
• Jalil Boukhobza (ENSTA Bretagne)
• Thierry Coupaye (Orange Labs)
• Erik Elmroth (UMEA)
• David Eyers (University of Otago)
• Paulo Ferreira (University of Oslo)
• Pradeeban Kathiravelu (University Alaska Anchorage)
• Remous-Aris Koutsiamanis (IMT Atlantique)
• Gabriel Luhasz (West University of Timisoara)
• Joan Manuel Marquès (Open University of Catalonia)
• Rolando Martins (University of Porto)
• Odorico Mendizabal (Universidade Federal do Rio Grande)
• Leandro Navarro (Polytechnic University of Catalonia)
• Ioan Petri (University of Cardiff)
• Guillaume Pierre (University of Rennes)
• Davy Preuveneers (Katholieke Universiteit Leuven)
• Kai Rannenberg (Goethe University)
• Laurent Réveillère (University of Bordeaux)
• Etienne Riviere (UC Louvain)
• Ivan Rodero (University of Utah)
• Valerio Schiavoni (University of Neuchatel)
• Mennan Selimi (South East European University)

Dates:

●     Paper submission, Oct 5th, 2023

●     Notification, Oct 23th, 2023

●     Camera Ready, Oct 27th, 2023

Submission page: here

Workshop and Submission format:

This will be the second edition of the workshop following last year. Opening speech from the organizers followed by a keynote.  The workshop will welcome both full and short papers (for demos and posters). Ending with an invited panel followed by a breakout session on new directions and applications for decentralized clouds (dependent on the number of accepted papers).

Submitted manuscripts should be structured as technical papers and may not exceed:

• six (6) single-spaced double-column pages,
• four (4) for short papers, including demos and posters

The page limits include all the content, including bibliography, appendix, etc.

Submitted papers must adhere to the formatting instructions of the ACM SIGPLAN style, which can be found on the ACM template page. The font size has to be set to 10pt.

Submissions can be made at the following link: https://middlewedge23.hotcrp.com/

The proceedings of MiddleWEdge 2022 are available at:  https://dl.acm.org/doi/proceedings/10.1145/3565385