SemMOB enables dynamic registration of sensors via mobile devices, search, and near real-time inference over sensor observations in ad-hoc mobile environments (e.g., re ghting). We demonstrate SemMOB in the context of an emergency response use case that requires automatic and dynamic registrations of sensor devices and annotation of sensor observations, decoding of latitude-longitude information in terms of human sensible names, fusion and abstraction of sensor values using background knowledge, and their visualization using mash-up.
Mobile devices are pervasive and increasingly becoming powerful (in terms of
memory, processing power, on-board sensors). They have access to two sources
of sensor information (a) internal sensors (embedded into mobile devices), and
(b) external sensors (present in the vicinity of mobile devices), both reporting
observations of the physical world. Only way to understand the physical world
from observations is to associate semantics with the observations by grounding
them in events in the physical world. For example, sensor observations about
temperature, CO2, and wavelength of light can be used to infer a re type event.
This can be used by the rst responders to decide on the type of extinguisher
to be used. SemMOB is a system that embodies such semantic processing of
observations from ad-hoc mobile sensors. It extends the capability of the
SemSOS [
The rest of paper is organized as follows. In Section 2, we describe the system architecture and review SemSOS and the reasoner. After we describe the nature of queries that can be issued to SemMOB and its visualization capabilities in Section 3, we conclude in Section 4.
SemMOB has two major components: (a) SemSOS and (b) Reasoning engine,
as shown in Figure 1. Mobile devices with an Android client registers with
SemSOS using the standard SOS register sensor request. There are internal
sensors such as GPS, gyroscope and camera, and external sensors such as activity,
temperature, and gaseous sensors, reporting observations to the mobile device via
bluetooth. All observations are sent to SemSOS and annotated using the
SSNXG ontology [
Mobile devices with first responders temapcetirvaittyu,rea,ngdas, gesture sensors
Gyroscope Temperature
GPS Heart rate
Pedometer
Knowledge
Base Stores sensor metadata and observations along with reasoned information Visualization of realtime observations and query interface
Reasons over sensor observations and sensor metadata
SSN-XG
SemSOS provides a semantic layer over Sensor Observation Service (SOS). SOS is a XML-based standardization of sensor data discovery and access on the web proposed by Semantic Web Enablement (SWE) group of Open Geospatial Consortium (OGC). SemSOS enhances SOS deployments to be semantically rich. SemSOS provides the same interface as SOS while it replaces the backend for processing semantically rich metadata associated with the sensors and observations using a semantic database.
In this work, we have further extended SemSOS to accommodate and integrate diverse mobile devices which enter the sensor network in an ad hoc manner and is required to be integrated. Each mobile device may have di erent capabilities and di erent sensors connected to it. SemMOB deals with the sensor network dynamism, sensor diversity, and near real-time reasoning over sensor metadata and its observations. 2.2
Reasoning Engine
The reasoner fetches newly inserted sensor metadata and sensor observations
from the knowledge base and inserts inferred triples to the knowledge base so
that it can be queried by semantically intelligent clients. Part of the ontology
for modeling sensor capability along with its location is shown in Figure 2.
Once a sensor registers with SemMOB, the reasoner uses latitude and longitude
of the sensor to obtain the named location from GeoNames [
Other reasoning procedures may use knowledge of poisonous gases, re
extinguishers, etc., to decide emergency operation related equipments and chemicals.
In the context of re ghting, we modeled the sensors (e.g., gaseous sensors) by
extending the SSN-XG ontology [
System Capabilities and Visualization The SemMOB system users can (1) monitor sensor observations in near realtime, and visualized on a map, (2) reason over sensor observations, and (3) query sensor metadata, sensor observations, and inferred knowledge. SemMOB provides unique capability of dynamic registration, search, reasoning, visualization, and querying of ad hoc mobile sensors in a uni ed framework. SemMOB query interface and the Android application is shown in Figure 3.
We demonstrated the functionalities of SemMOB including dynamic registration, querying, reasoning, and visualization of observations from mobile ad-hoc devices and sensors attached to them.