![](bg2.png)
pedestrians usually walk on sidewalks and not on
road center lines. For effective navigation assistance
to pedestrians, PNSs must contain pedestrian paths
in the underlying navigation environments. The core
data in a pedestrian path database is a network of
sidewalks which contains geometry, topology, and
attributes of the pedestrian path (Kasemsuppakorn
and Karimi, 2009). Geometrical data in a pedestrian
path database include coordinates of decision
points—a decision point is a juncture whereby a
decision must be made by the pedestrian (e.g., an
intersection)—and of intermediate points on
sidewalks. Topological data in a pedestrian path
database include data about connectivity between
decision points and sidewalks. Attribute data in a
pedestrian path database includes data such as
sidewalk name, sidewalk width, and sidewalk type.
Given all these PNS components and features,
the limitations of mobile devices, diversity of user
requirements, and the environmental factors, a
dynamic configuration scheme is a key requirement
for providing efficient services. For example, given
a complex computational task and a mobile device
with low battery power, it is best to have the task
performed on the server side. On the other hand, a
simple computational task that is required frequently
is best to be performed in the client side to minimize
communication overheads. Many combinations,
considering device constraints and computations at
hand, among other factors, are possible for which an
efficient solution is desirable.
The contribution of the paper is an algorithm for
automatic configuration of data and computation
components in PNSs. The algorithm addresses many
situations where minimizing energy consumption is
the highest priority. The rest of the paper is
organized as follows. In Section 2, the main
components of PNSs and possible model
configurations are discussed. Section 3 describes the
algorithm. The models and the simulation of the
algorithm are explained in Section 4. Simulation
results are discussed in Section 5. Section 6
highlights related works and Section 7 concludes the
paper and discusses future research.
2 COMPONENTS AND MODELS
The main components and functions of PNSs
include sidewalk network map database, geocoding,
map matching, routing, smartphone (GPS-enabled),
and cloud computing server and/or service provider.
Sidewalk network map databases are the main data
source for PNSs. Geocoding is the process of
assigning geographic coordinates (lat/long) to a
given place by comparing its description to the
descriptions of location-specific elements in the
reference datasets (Goldberg, 2007). Map matching
is a technique for matching pedestrian trajectories to
correct pedestrian paths in the sidewalk network
(Karimi et al., 2006). Routing computes a preferred
path from any given origin to destination
(Kasemsuppakorn and Karimi, 2009). Proposed
models of PNSs are explained bellow in brief.
2.1 Minimum Computation (MinComp)
For a resource poor client device, the MinComp
model is considered where most computations are
submitted to the cloud. In the MiniComp model,
with the assumption that the device battery is low,
its computation capability is limited, and its
available storage is low, all computational tasks are
submitted to the cloud. In this model, the client is
mainly responsible for submitting the user request to
the cloud, receiving the response from the cloud
upon task completion, and presenting the results to
the user.
2.2 Minimum Communication
(MinComm)
In the MinComm model, most computations are
performed at the client device provided that the
client has sufficient battery power, computational
capability, and storage capacity. This model is
suitable in situations where the device can handle
the computational load and the user needs fast
response. This model is also suitable in situations
where network availability is limited and network
quality is poor. The main responsibilities of the
cloud in this model are map storage, map rendering,
and map data transfer.
2.3 Balanced Computation-
communication (BalCC)
The third model, BalCC, is the most suitable model
for many clients since client devices fall into this
category most often. This is basically a trade-off
between the MinComp and MinComm models. In
situations where the client device has an average
battery power, average computation capability,
moderate amount of storage capacity, and network
availability is in moderate level, the BalCC model
provides the most optimal solution. In this model,
some of the computations are performed in the client
side while some in the cloud maintaining a balance
SMARTGREENS2012-1stInternationalConferenceonSmartGridsandGreenITSystems
86