31-01-2013, 04:34 PM
Detecting Hidden and Exposed Terminal Problems in Densely DeployedWireless Networks
1Detecting Hidden.pdf (Size: 994.61 KB / Downloads: 25)
Abstract
In this paper, we discuss problems in densely deployed
wireless networks. Particularly, we focus on wireless local
area networks (WLANs) because they have enabled us to provide
seamless and high capacity wireless access easily and inexpensively.
However, recently, channel interference between different
services has become a serious problem because access points
(APs) of WLANs are located too densely. In the carrier sense
multiple access with collision avoidance used in WLANs, the
hidden terminal (HT) and the exposed terminal (ET) problems
occur depending on the distance between stations and the carrier
sensing range. In the higher dense deployment mentioned above,
the HT and ET problems occur complicatedly. Therefore, we
propose an AP cooperation system that detects the HT and
ET problems between stations (STAs). In our system, APs are
operated with time synchronization and obtain the information of
connected STAs from received frames. The HT and ET problems
are identified from the integration of the information obtained
at different APs. The effectiveness is verified by simulations.
INTRODUCTION
THE extensive use of IEEE 802.11 wireless local networks
(WLANs) has been brought, thanks to their fast
communication speed, license free operation and inexpensive
deployment. In fact, they are widely used in homes, offices,
and public spaces as part of our social infrastructure. However,
because each service provider or each personal user has
located his or her basic service set (BSS), which consists
of an access point (AP) and one or more stations (STAs),
distributedly and selfishly without taking into account other
preexisting BSSs, interference between BSSs that are deployed
too densely has become a serious problem. It is reported in
[1] that a BSS is interfered with by eighty other BSSs in the
worst case scenario observed in San Francisco. In such a dense
WLAN environment, wireless bandwidth is used inefficiently
due to frequent frame collisions and unnecessary suppression
of transmission; the carrier sense multiple access with collision
avoidance (CSMA/CA) protocol does not work properly.
RELATED WORK
Since the majority of existing works about the MAC-level
problems have focused on mitigating the effects of problems,
the detection of the problems remains ill-argued. Especially,
it has been thought that the ET occurrences are harder to
detect than the HT because the only way to detect is by
disabling carrier sensing at the STAs and testing for loss-free
simultaneous communication. In [8], the authors have proposed
Mutual Observation with Joint Optimization (MOJO) to
detect the HT problem between STAs in a single BSS. Choi
proposed a scheme that an AP collects the carrier sensing
information among the STAs associated with it and detects
HT problems[9]. However, these schemes do not take into
account the ET problem. Moreover, the detection accuracy has
not been validated quantitatively. Alternatively, several works
have tried to mitigate the effect of MAC-level problems by the
indirect detection of observing the interference and collisions.
The authors in [11] design an AP selection strategy by using
the potential HT problems. However, all the above schemes
require large modifications from the existing STAs.
MAC-LEVEL PROBLEMS IN DENSELY DEPLOYED
WLANS
The IEEE 802.11 WLAN adopts the distributed coordination
function based on CSMA/CA as a basic medium access
protocol, where each STA distributedly determines the timing
of transmission of its frames. In CSMA/CA, to avoid frame
collisions, each STA senses if the medium is being used by
another STA before it starts transmission. However, improper
carrier sensing results in frequent frame collisions between
STAs and excessive suppression of transmitting frames, which
leads to the waste of bandwidth.
Detection system
Our system integrates information observed at APs using
the same channel on the database and then estimates κij , λij ,
and μij defined above from the database. The basic procedure
of our system is simply described as in Fig. 4 and follows:
i) APs enter the predetermined observation period; ii) during
the observation period, each AP obtains the information, i.e.,
MAC address, time stamp, and received power from data
frames that it received from STAs; iii) after the observation
period, our system integrates the information observed at APs
and identifies the MAC-level problems from the integrated
information.
SIMULATION DESCRIPTION
Simulations were carried out to validate the effectiveness
of our system using the topology shown in Fig. 5. We fixed
the field size and the number of STAs in the simulation, while
we varied the AP interval which is the distance between APs.
Here, we consider the scenario that three WLAN services are
mixed in the area; each AP is managed by one of the service
providers, while each STA is connected to the AP with the
highest RSS among only the APs provided by the service it
subscribes to. This scenario allows us to reproduce various
MAC-level problems illustrated in Fig. 3. We observed APs
and STAs in the three services using a particular channel (1).
APs belonging to different services can cooperate to detect
the MAC-level problems if they use the same frequency band
CONCLUSION
In this paper, we focused on the uplink and categorized the
MAC-level problems that occur in densely deployed WLANs:
a) HT problem in a single BSS, b-1) HT problem between
different BSSs, b-2) HT problem due to superimposed power,
c-1) ET problem, and c-2) ET problem due to superimposed
power. Next, we proposed a detection system particularly for
problems a), b-1), and c-1). In our system, APs using the same
channel cooperatively share their observed information. Our
system integrates the observed information and then detects
the MAC-level problems based on their connected APs, their
carrier sense relationship, and their frame-collision possibility,
which are estimated from the integrated information. Our
simulation results showed that our system can detect the
MAC-level problems accurately, regardless of AP density or
shadowing effect. In the future, in addition to the remaining
issues in Section VII, we have to investigate the detection of
MAC-level problems in the downlink.