Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Dynamic Update of CSP Allocations in Federation Orchestration

Beke, Marieke LU (2024) EITM01 20232
Department of Electrical and Information Technology
Abstract
This paper discusses the project in which an algorithm was written to dynamically
approach federation orchestration. This builds on and is part of the REINDEER
project. In an environment with goods and online agents, and in this specific case,
Contact Service Points (CSPs) and User Equipments (UEs), where the agents have
different applications and make unpredictable changes such as moving around a
room, finding a dynamic reallocation algorithm is a significant challenge.

The use of diverse case studies, encompassing both minor and significant adjust-
ments to an allocation, has proven to be an effective approach in identifying more
optimal solutions. An iterative methodology has been employed, wherein a ran-
domly selected case... (More)
This paper discusses the project in which an algorithm was written to dynamically
approach federation orchestration. This builds on and is part of the REINDEER
project. In an environment with goods and online agents, and in this specific case,
Contact Service Points (CSPs) and User Equipments (UEs), where the agents have
different applications and make unpredictable changes such as moving around a
room, finding a dynamic reallocation algorithm is a significant challenge.

The use of diverse case studies, encompassing both minor and significant adjust-
ments to an allocation, has proven to be an effective approach in identifying more
optimal solutions. An iterative methodology has been employed, wherein a ran-
domly selected case is evaluated in each iteration. Only if a superior utility value
is achieved is the new solution accepted. The utility values are calculated by a
entity, designated as the oracle, which was not a part of the original project scope.

This approach has yielded higher, yet more consistent utility values compared to
random reallocation. This is the case for different numbers of iterations, different
numbers of CSPs and different numbers of UEs.

The runtime can become a bottleneck of the algorithm if more iterations have to
be chosen. This is due to the fact that the oracle must be called every iteration.
Overall, the addition of the dynamic reallocation algorithm is a relevant addition. (Less)
Popular Abstract
In today’s rapidly evolving technological landscape, efficient resource
management is crucial for the optimal performance of distributed sys-
tems. This project, part of the innovative REINDEER research project,
focuses on dynamic resource allocation within federated networks, where
multiple independent entities collaborate to share resources like pro-
cessing power, storage, and connectivity. These networks must swiftly
adapt to the unpredictable movements and varying demands of users.
The intelligent algorithm continuously adjusts resource distribution,
based on data input, saving computing time and improving efficiency.
This research has implications for the future of networked systems,
enabling technologies that support... (More)
In today’s rapidly evolving technological landscape, efficient resource
management is crucial for the optimal performance of distributed sys-
tems. This project, part of the innovative REINDEER research project,
focuses on dynamic resource allocation within federated networks, where
multiple independent entities collaborate to share resources like pro-
cessing power, storage, and connectivity. These networks must swiftly
adapt to the unpredictable movements and varying demands of users.
The intelligent algorithm continuously adjusts resource distribution,
based on data input, saving computing time and improving efficiency.
This research has implications for the future of networked systems,
enabling technologies that support next-generation communication in-
frastructure.

In the realm of federated environments,
efficient resource management is pivotal
for optimal performance. The REIN-
DEER project, funded by the European
Union’s Horizon 2020 programme, aims
to revolutionize connectivity through
RadioWeaves technology, focusing on
resilience, interactivity, hyper-diversity,
and energy efficiency.

A crucial part of the RadioWeaves in-
frastructure is federation orchestration,
this implies managing Contact Service
Points (CSPs) and User Equipments
(UEs). CSPs are the resources, indi-
visible goods that serve the UEs, on-
line agents. UEs with similar applica-
tions are grouped into federations to
streamline resource allocation. The key
challenge addressed in this project is
the development of a dynamic reallo-
cation algorithm. Unlike traditional
static methods, this algorithm adapts
to changes such as UEs coming online
or moving within a room, ensuring op-
timal CSP allocation. It leverages an
iterative approach where adjustments
are made incrementally based on utility
evaluations by an evaluation compo-
nent, the oracle.

Testing and analysis demonstrate the
algorithm’s effectiveness across varying
scenarios, including different numbers
of CSPs and UEs. Results consistently
show improved utility values compared
to static allocation methods, confirm-
ing its suitability for dynamic environ-
ments.

Looking ahead, future enhancements
could relax constraints on application
homogeneity among UEs within feder-
ations, explore weighted case prioriti-
sation, and optimise computational ef-
ficiency to further refine allocation re-
sults. These efforts will pave the way
for more efficient and adaptable network
infrastructures in the future. (Less)
Please use this url to cite or link to this publication:
author
Beke, Marieke LU
supervisor
organization
course
EITM01 20232
year
type
H2 - Master's Degree (Two Years)
subject
report number
LU/LTH-EIT 2024-993
language
English
id
9166279
date added to LUP
2024-06-20 14:52:45
date last changed
2024-06-20 14:52:45
@misc{9166279,
  abstract     = {{This paper discusses the project in which an algorithm was written to dynamically
approach federation orchestration. This builds on and is part of the REINDEER
project. In an environment with goods and online agents, and in this specific case,
Contact Service Points (CSPs) and User Equipments (UEs), where the agents have
different applications and make unpredictable changes such as moving around a
room, finding a dynamic reallocation algorithm is a significant challenge.

The use of diverse case studies, encompassing both minor and significant adjust-
ments to an allocation, has proven to be an effective approach in identifying more
optimal solutions. An iterative methodology has been employed, wherein a ran-
domly selected case is evaluated in each iteration. Only if a superior utility value
is achieved is the new solution accepted. The utility values are calculated by a
entity, designated as the oracle, which was not a part of the original project scope.

This approach has yielded higher, yet more consistent utility values compared to
random reallocation. This is the case for different numbers of iterations, different
numbers of CSPs and different numbers of UEs.

The runtime can become a bottleneck of the algorithm if more iterations have to
be chosen. This is due to the fact that the oracle must be called every iteration.
Overall, the addition of the dynamic reallocation algorithm is a relevant addition.}},
  author       = {{Beke, Marieke}},
  language     = {{eng}},
  note         = {{Student Paper}},
  title        = {{Dynamic Update of CSP Allocations in Federation Orchestration}},
  year         = {{2024}},
}