ACCESS project
COoperative COmmunication in Multihop wireless networks
COCOM
Project leaders:
Viktoria Fodor, faculty (LCN),
Ragnar Thobaben, faculty (CT)
Project members:
Mikael Skoglund, founding faculty (CT)
Lars Rasmussen, founding faculty (CT)
Mikael Johansson, founding faculty (AC)
Liping Wang, Ph.D. student (LCN)
Nicolas Schrammar, Ph.D. student (CT)
Maksym Girnyk, Ph.D. student (CT)
Objectives
Efficient communication in multihop wireless networks depends on several factors: on the
physical topology of the network, with links between nodes that can decode each others
transmission, on the logical topology, with links between nodes that supposed to listen to each
others transmission, on the routing algorithm that determines how to forward information through
multiple hops, on the scheduling and channel access methods that control how shared radio
resources are accessed, and on physical layer methods including channel coding and advanced
techniques of cooperative communication and interference control.
While all these issues were traditionally addressed separately, the advance of technology and
theory on various fields — from complex systems theory to communication theory, from flexible
power control to increasing node processing power — makes it possible to significantly improve
the performance and reliability of multihop wireless networks by optimizing physical, medium
access and network control functions together.
The goal of this project is two-fold. First, we aim at defining cross-layer models and design
methodologies that allow tractable system design in this complex networking scenario; second,
we plan to address practical network design problems of, for example, multihop cognitive radio
networks and industrial control networks.
This project is a continuation of two tasks in earlier ACCESS projects: one considered the
problem of efficient communication in sensor networks for distributed sensing, and combined
results of cooperative communication and networking, the second considered topology control for
energy limited wireless networks, building efficient topologies by using methods of classical
networking and complex system theory.
Deliverables
Deliverable 1. Capacity limits of multihop relay networks
The evaluation of cooperative communication techniques often builds on an information theoretic
approach, which makes it difficult to extend the results for complex networking scenarios. The
deterministic model, proposed recently by Tse et. al., provides a tool that brings communication
theory and networking closer to each other and allows the use of simplified graph models to
describe the capabilities of cooperative schemes. We will evaluate the accuracy of the
deterministic model in some typical scenarios and use the models to derive the capacity of
multiple networks under co-optimized logical network topology design and relaying scheme.
Deliverable 2. Distributed channel coding and routing for multihop relay networks
The networking environment of wireless networks changes continuously due to the effect of
interference from the network nodes in the same network of from networks located in the same
area, due to the eventual mobility of the nodes, or changes in the radio propagation
characteristics. The dynamic networking environment requires adaptive network control functions
that are able to change the way network resources are utilized. In this work we will jointly
consider routing and channel coding, two control functions, which are both very sensitive to the
changes of the environment and affect the quality of parallel transmissions in the network, in
order to achieve an improved performance with a cross-layer design. Specifically, we will
consider a multihop network, where the channel conditions change dynamically, and propose a
distributed routing and channel coding algorithm that builds the transmission path and selects the
cooperative coding scheme hop-by-hop, according to the quality of data available at the network
nodes.
Deliverable 3. Resilient routing under delay constraints
A specific feature of industrial communications is the combination of non-saturated and
correlated traffic with (often stringent) real-time constraints on individual packet delivery. The
scheduling and routing of such traffic is fundamentally different from the scheduling problems
for saturated sources (i.e., where sources always have a new packet to send) that have been
studied for a long time in the context of packet radio networks. Moreover, since the wireless
medium is shared and since typical wireless sensor nodes often use half-duplex radios, scheduling
policies for individual nodes cannot be studied in isolation but have to be coordinated across the
network and ideally integrated with routing decisions. Since transmission errors are inherent in
low-power wireless communication (especially in a harsh and dynamic industrial environment)
an important principle for increased system reliability multi-hop mesh networking combined with
resilient routing. We plan to develop techniques for computing “full diversity routes,” which
explore the full time-, space-, coding- and frequency-diversity available to multi-hop wireless
sensor and actuator networks. We will also explore synergies with the multihop relay networks
described in Deliverable 2; essentially, relaying is a physical layer technology while resilient
routing, as discussed here, sits at the network layer.
Deliverable 4. Joint physical-layer and network-layer coding in multihop relay networks
Decentralized network coding, suitable for wireless networks, can be based on random linear
network codes where each intermediate node creates random linear combinations of incoming
packets. Wireless channels exhibit inherent features resembling the encoding operation of nodes
that perform random linear network coding: due to the broadcast and multiple-access features of
wireless transmission, each node receives a linear combination of transmitted signals, weighting
by random channel coefficients. This inherent encoding capability has potential to support
elaborate joint physical-layer and network-layer coding strategies, seamlessly operating across a
multi-hop wireless network. In this work we will investigate such strategies based on a crosslayer perspective. The overall aim is to develop an analytical framework for performance
evaluation and for identifying constraints and technology challenges for realizing wireless
random linear network coding in multihop networks.