11-07-2012, 03:15 PM
Continuous Neighbor Discovery in Asynchronous
Sensor Networks
Continuous Neighbor Discovery in Asynchronous.pdf (Size: 431.59 KB / Downloads: 42)
INTRODUCTION
ASENSOR network may contain a huge number of simple
sensor nodes that are deployed at some inspected site. In
large areas, such a network usually has a mesh structure. In this
case, some of the sensor nodes act as routers, forwarding messages
from one of their neighbors to another. The nodes are configured
to turn their communication hardware on and off to minimize
energy consumption. Therefore, in order for two neighboring
sensors to communicate, both must be in active mode.
In the sensor network model considered in this paper, the
nodes are placed randomly over the area of interest, and their
first step is to detect their immediate neighbors—the nodes with
which they have a direct wireless communication—and to establish
routes to the gateway. In networks with continuously heavy
traffic, the sensors need not invoke any special neighbor discovery
protocol during normal operation.
RELATED WORK
In a WiFi network operating in centralized mode, a special
node, called an access point, coordinates access to the shared
medium. Messages are transmitted only to or from the access
point. Therefore, neighbor discovery is the process of having a
new node detected by the base station. Since energy consumption
is not a concern for the base station, discovering new nodes
is rather easy. The base station periodically broadcasts a special
HELLO message.1 A regular node that hears this message
can initiate a registration process. The regular node can switch
frequencies/channels in order to find the best HELLO message
for its needs. Which message is the best might depend on the
identity of the broadcasting base station, on security considerations,
or on PHY layer quality (signal-to-noise ratio).
BASIC SCHEME AND PROBLEM DEFINITION
In the following discussion, two nodes are said to be neighboring
nodes if they have direct wireless connectivity. We assume
that all nodes have the same transmission range, which
means that connectivity is always bidirectional. During some
parts of our analysis, we also assume that the network is a unit
disk graph; namely, any pair of nodes that are within transmission
range are neighboring nodes. Two nodes are said to be
directly connected if they have discovered each other and are
aware of each other’s wake-up times. Two nodes are said to be
connected if there is a path of directly connected nodes between
them. A set of connected nodes is referred to as a segment. Consider
a pair of neighboring nodes that belong to the same segment
but are not aware that they have direct wireless connectivity.
See, for example, nodes and in Fig. 4(a).
ESTIMATING THE IN-SEGMENT DEGREE OF A HIDDEN NEIGHBOR
As already explained, we consider the discovery of hidden
neighbors as a joint task to be performed by all segment nodes.
To determine the discovery load to be imposed on every segment
node, namely, how often such a node should become active
and send HELLO messages, we need to estimate the number
of in-segment neighbors of every hidden node , denoted by
. In this section, we present methods that can be used
by node in the Normal (continuous neighbor discovery) state
to estimate this value. Node is assumed to not yet be connected
to the segment, and it is in the Init (initial neighbor discovery)
state. Three methods are presented.
CONCLUSION
We exposed a new problem in wireless sensor networks, referred
to as ongoing continuous neighbor discovery. We argue
that continuous neighbor discovery is crucial even if the sensor
nodes are static. If the nodes in a connected segment work together
on this task, hidden nodes are guaranteed to be detected
within a certain probability and a certain time period , with
reduced expended on the detection.
Sensor Networks
Continuous Neighbor Discovery in Asynchronous.pdf (Size: 431.59 KB / Downloads: 42)
INTRODUCTION
ASENSOR network may contain a huge number of simple
sensor nodes that are deployed at some inspected site. In
large areas, such a network usually has a mesh structure. In this
case, some of the sensor nodes act as routers, forwarding messages
from one of their neighbors to another. The nodes are configured
to turn their communication hardware on and off to minimize
energy consumption. Therefore, in order for two neighboring
sensors to communicate, both must be in active mode.
In the sensor network model considered in this paper, the
nodes are placed randomly over the area of interest, and their
first step is to detect their immediate neighbors—the nodes with
which they have a direct wireless communication—and to establish
routes to the gateway. In networks with continuously heavy
traffic, the sensors need not invoke any special neighbor discovery
protocol during normal operation.
RELATED WORK
In a WiFi network operating in centralized mode, a special
node, called an access point, coordinates access to the shared
medium. Messages are transmitted only to or from the access
point. Therefore, neighbor discovery is the process of having a
new node detected by the base station. Since energy consumption
is not a concern for the base station, discovering new nodes
is rather easy. The base station periodically broadcasts a special
HELLO message.1 A regular node that hears this message
can initiate a registration process. The regular node can switch
frequencies/channels in order to find the best HELLO message
for its needs. Which message is the best might depend on the
identity of the broadcasting base station, on security considerations,
or on PHY layer quality (signal-to-noise ratio).
BASIC SCHEME AND PROBLEM DEFINITION
In the following discussion, two nodes are said to be neighboring
nodes if they have direct wireless connectivity. We assume
that all nodes have the same transmission range, which
means that connectivity is always bidirectional. During some
parts of our analysis, we also assume that the network is a unit
disk graph; namely, any pair of nodes that are within transmission
range are neighboring nodes. Two nodes are said to be
directly connected if they have discovered each other and are
aware of each other’s wake-up times. Two nodes are said to be
connected if there is a path of directly connected nodes between
them. A set of connected nodes is referred to as a segment. Consider
a pair of neighboring nodes that belong to the same segment
but are not aware that they have direct wireless connectivity.
See, for example, nodes and in Fig. 4(a).
ESTIMATING THE IN-SEGMENT DEGREE OF A HIDDEN NEIGHBOR
As already explained, we consider the discovery of hidden
neighbors as a joint task to be performed by all segment nodes.
To determine the discovery load to be imposed on every segment
node, namely, how often such a node should become active
and send HELLO messages, we need to estimate the number
of in-segment neighbors of every hidden node , denoted by
. In this section, we present methods that can be used
by node in the Normal (continuous neighbor discovery) state
to estimate this value. Node is assumed to not yet be connected
to the segment, and it is in the Init (initial neighbor discovery)
state. Three methods are presented.
CONCLUSION
We exposed a new problem in wireless sensor networks, referred
to as ongoing continuous neighbor discovery. We argue
that continuous neighbor discovery is crucial even if the sensor
nodes are static. If the nodes in a connected segment work together
on this task, hidden nodes are guaranteed to be detected
within a certain probability and a certain time period , with
reduced expended on the detection.