Multi-Objective Scheduling for MUD based
Ad-Hoc Networks
M. Bouharras Z.Dziong F. Gagnon M. Haidar
Department of Electrical Engineering,
Ecole de Technologie Supérieure
Montreal, Quebec, Canada, H3C 1K3
(Mohamed.bouharras.1, Mohamad.Haidar.1}@ens.etsmtl.ca, {Zbigniew.Dziong, Francois.Gagnon}@etsmtl.ca
ABSTRACT
Common channel multi-hop Ad Hoc networks have
some inherent constraints related to throughput and
Quality of Service (QoS). Multiuser detection (MUD)
based Medium Access Control (MAC) can relax some of
these constraints and provide significant gains in
throughput and Quality of Service (QoS). These gains can
be realized by implementing a distributed neighborhood
scheduling algorithm that needs to choose one from
several possible transmission configurations in each
frame. This feature allows formulating different
scheduling performance objectives such as delay
minimization or throughput maximization. In this paper
we focus on analysis and comparison of the system
performance under different objectives including multi-
objective formulations. First we implement a scheduling
scheme that minimize delay using Start Time Fair
Queuing (STFQ) algorithm and compare its performance
with scheduling that maximises the throughput. Then we
formulate multi-objective functions that are used to
achieve a trade-off between delay and throughput
performance. One of these formulations is based on the
Nash arbitration scheme from cooperative game theory.
The numerical results demonstrate the flexibility and
efficiency of the proposed approach.
Keywords
Multiuser detection; Multiuser reception; MAC protocol;
scheduling; fairness and wireless ad hoc networks.
1. INTRODUCTION
Wireless ad hoc networks are expected to support
multimedia applications in emergency disaster
management and military operations. This class of
mission-critical applications demands a certain level of
quality of services (QoS) for proper operation. Many
MAC protocols have been developed for wireless
networks which are based on a common channel shared
by mobile hosts in the network. These protocols, such as
IEEE 802.11 are referred to as single-channel protocols.
Due to relatively high probability of contentions and
collisions, the performance of single-channel MAC
protocols deteriorates quickly while the number of mobile
hosts increases, especially for muli-hop connections [1].
To mitigate this problem, one can consider utilizing
multiuser detection (MUD) or multiuser reception (MUR)
CDMA technology that allows the reception of multiple
CDMA channels at the same time. In [2] and [3] the
authors proposed a novel MAC and scheduling algorithms
to take advantage of MUD and MUR. One of the key
elements of their approach is a distributed neighborhood
scheduling mechanism that is based on a protocol that
exchanges information between neighbors. They
presented numerical results indicating significant gains in
the QoS (delay) and network throughput. The objective of
their scheduling mechanism was based on a principle that
priority was given to the voice packets that waited the
longest time in the neighborhood’s queues. This principle
is not necessary optimal as shown in the studies of fair
queuing mechanisms. Also, the network operation can
have multi-objective formulation where the scheduling
should balance network throughput maximization
objective with traffic class fair access optimization.
In this paper, we first propose an approach for
analyzing and comparing close to optimal distributed
neighborhood scheduling schemes with different
objectives. The approach is based on the notion of a flow
contention graph that takes into account the topology of
the network. Using this notion, we construct a dependence
matrix of flows in each node that constitutes a base for
selecting optimal configuration of transmissions in a
given slot according to the chosen objective.
Then, using the above platform we implement several
scheduling mechanisms. We start with single objective
schemes such as scheduling that minimize delay using
Start Time Fair Queuing (STFQ) algorithm and
scheduling that maximizes the throughput. Then we
formulate multi-objective functions that are used to
achieve a trade-off between delay and throughput
performance. Two concepts are considered. The first one
is based on the Nash arbitration scheme from cooperative
game theory where the objective is to maximize the
product of two utility functions. The second one is based
on maximization of the weighted sum of these utility
functions.
The numerical results demonstrate the viability of the
proposed scheduling platform. First we compare the delay
and throughput performance of the scheduling scheme
that utilizes Start Time Fair Queuing (STFQ) algorithm
WICON'08, November 17-19, 2008, Maui, Hawaii, USA.
Copyright 2008 ICST 978-963-9799-36-3
1