Middleware for Sensor Network

1
Middleware for Sensor Network

• ECE1770 Trends in Middleware Systems
• Prepared by Mi Li and Tony Lee

2
Agenda

• 1. Introduction of Sensor Network
• 1.1 Application Examples
• 1.2 Sensor Network Architecture
• 1.3 Hardware and Software
• 1.4 Sensor Network's Features
• 2. Middleware for Sensor Network.
• 2.1 Why do we need Middleware for SN?
• 2.2 Challenges in designing middleware.
• 2.3 Middleware approaches for SN.
• 2.4 Example Mires, A Message-Oriented Middleware

3
1.1 Application Examples

• Military applications
• target detection
• battlefield surveillance
• nuclear, biological and chemical attack detection
  etc.
• Enviromental applications
• forest fire detection
• flood detection
• structure monitoring
• Health applications
• monitor human physiological data
• drug administration

4
(No Transcript)
5
(No Transcript)
6
Agenda

• 1. Introduction of Sensor Network
• 1.1 Application Examples
• 1.2 Sensor Network Architecture
• 1.3 Hardware and Software
• 1.4 Sensor Network's Features
• 2. Middleware for Sensor Network.
• 2.1 Why do we need Middleware for SN?
• 2.2 Challenges in designing middleware.
• 2.3 Middleware approaches for SN.
• 2.4 Example Mires, A Message-Oriented Middleware

7
1.2 Sensor Network Architecture

• Sensor Nodessense target events, gather sensor
  readings, manipulate informations, send them to
  gateway via radio link
• Base station/sink communicate with sensor nodes
  and user/operator, (database-stores the data)
• Operator/user task manager, send query

8
1.2 Sensor Network Architecutre cont.
Task Management Plane
Mobility Management Plane
Application Layer
Transport Layer
Power Management Plane
Network Layer
Data Link Layer
Application Layer middleware, OS Network Layer
Routing
Physical Layer
9
1.2 Sensor Network Architecture-Routing

• State of the art routing protocols are
  distributed and reactive the systems start
  looking for a route only when they have
  application data to transmit
• We study here Ad hoc On demand Distance Vector
  (AODV) and Dynamic Source Routing (DSR) for the
  sensor network

10
1.2 Sensor Network Architecture-Routing cont.

• Route Discovery
• A node sends a Route Request message to all of
  its neighbours.
• Any node receiving such a request, either
  answers to it or rebroadcasts it.
• The procedure finishes either when the
  request sender has received the
• route information, or when the request
  times out.
• With AODV, each node remembers the next hop
  information associated with the destination. The
  route knowledge itself is distributed in the
  network.
• With DSR, the complete route is sent to the route
  requester.
• Message transmission
• With AODV, the message is sent to the next hop as
  recorded in the routing table, and this procedure
  is repeated at each hop.
• With DSR, the message is sent with its complete
  route as header.

11
1.2 Sensor Network Architecture- Routing cont.

• Rumor Routing
• "Rumor Routing Algorithm for Sensor Network" by
  Braginsky and Estrin
• How to make information available in a sensor
  network
• Assumption sense particular eventt when
  requested, don't know the existence or the
  location of the event

12
1.2 Sensor Network Architecture- Routing cont.

• An event sends out agents which travel the
  network from node to node on a random path.Each
  visit leaves information about the event in the
  node's database. After a predefined TTL the agent
  stops
• A requester also sends out an agent. After some
  time it will hopefully come across the path of
  the information agent by checking the node's
  databases. It can then travel the backward
  references the first agents left in the nodes to
  reach the event.

13
1.2 Sensor Network Architecture- Routing cont.

• Critical review
• Only a small number of nodes have to adopt
  the same information
• Only a small number of nodes have to process
  the request When or whether requested
  information can be delivered is a random process.
• - The failure of nodes can interrupt the path
  to the event (depending on how broad it is).
• - The actual behavior of a node is very
  different from what is shown in the former slides

14
Agenda

• 1. Introduction of Sensor Network
• 1.1 Application Examples
• 1.2 Sensor Network Architecture
• 1.3 Hardware and Software
• 1.4 Sensor Network's Features
• 2. Middleware for Sensor Network.
• 2.1 Why do we need Middleware for SN?
• 2.2 Challenges in designing middleware.
• 2.3 Middleware approaches for SN.
• 2.4 Example Mires, A Message-Oriented Middleware

15
1.3 Hardware - Sensor Nodes

• Sensing sensor --a transducer that converts a
  physical, chemical, or biological parameter into
  an electrical signal
• Processing microprocessor(CPU)
• data storage(Mem)
• AD converter
• Communicating data transceiver(Radio),
• Energy source battery

16
Hardware - Sensor Nodes cont.
17
Hardware Sensor Nodes cont.
18
Hardware - Sensor Nodes cont.
19
Hardware Sensor Nodes cont.

• Characteristics of Sensor Nodes
• Limited capacity of
• Battery (Lifetime day - 10 years)
• Processing capabilities (10MHz)
• Transmission range (5 - 20 meters)
• Data rates Bit/s - KB/s
• Transmission methods
• 802.11 (WiFi)
• Bluetooth short distance, other applications
• ZigBee for sensor network
• Price some cents

20
Software - Operating System

• Sensor Manager
• provides access to the sensors
• manage the delivery of sensor data
• provides and manages the resources provided to a
  sensor

Integrity Service/ Access Control
Query Manager
Storage
Sensor Manager
21
Software - Operation System cont.
Integrity Service/ Access Control
Query Manager
Storage
Sensor Manager

• Storage
• persistent storage for data streams

22
Software - Operating System cont.

• Query Manager
• manages active queries
• query processing
• delivery of events and query results to
  registered, local or remote consumers

Integrity Service/ Access Control
Query Manager
Storage
Sensor Manager
23
Software - Operation System cont.

• Top layer access control and integrity service
• OS examples
• TinyOS when an event occurs, it calls the
  appropriate event handler to handle the event.
• Others Contiki, MANTIS, and SOS.

Integrity Service/ Access Control
Query Manager
Storage
Sensor Manager
24
Software Design

• Create Hardware-optimized software components
  (driver, operating system )
• Create hardware- independent software components
  (middleware, services)
• Combining of predefined components
• Source code generation
• Removing unused components
• Optimizaion of interface
• Optimizaion to node's hardware
• Distribution of nodes in different environments
• Monitoring the execution
• Creation of logfiles
• Evaluation of logfiles

25
Agenda

• 1. Introduction of Sensor Network
• 1.1 Application Examples
• 1.2 Sensor Network Architecture
• 1.3 Hardware and Software
• 1.4 Sensor Network's Features
• 2. Middleware for Sensor Network.
• 2.1 Why do we need Middleware for SN?
• 2.2 Challenges in designing middleware.
• 2.3 Middleware approaches for SN.
• 2.4 Example Mires, A Message-Oriented Middleware

26
1.4 Sensor Network's features

• Restricted Resources energy, computing power,
  memory, and communication bandwidth
• Network Dynamics node mobility, environmental
  obstructions, hardware failures
• Scale of Deployments thousands or millions of
  nodes

27
Sensor Network's features cont.

• Real-world intergation seperation of events in
  time and space and correlate information from
  multiple sources
• Collection and Processing of Sensor Data
  preprocess data at the source
• e.g. smart sensor ( IEEE 1451 )
• Integration with Background Infrastructures
  tasking the sensor network, and provide resources
  (eg. computing power)

28
Agenda

• 1. Introduction of Sensor Network
• 1.1 Application Examples
• 1.2 Sensor Network Architecture
• 1.3 Hardware and Software
• 1.4 Sensor Network's Features
• 2. Middleware for Sensor Network.
• 2.1 Why do we need Middleware for SN?
• 2.2 Challenges in designing middleware.
• 2.3 Middleware approaches for SN.
• 2.4 Example Mires, A Message-Oriented Middleware

29
2.1 Why do we need Middleware for WSN?

• What is it?
• A software infrastructure that glues together the
  network hardware, operating systems, network
  stacks, and applications.
• Role
• Provide standardized system services to diverse
  applications.
• Provide a runtime environment that can support
  and coordinate multiple applications.
• Provide mechanisms to achieve adaptive and
  efficient utilization of system resources.

30
Compare with other middlewares.

• Traditional middleware systems
• Normally heavyweight in terms of memory and
  computation and therefore not suitable for SNs.
• Java RMI (Remote Method Invocation)
• EJB (Enterprise JavaBeans)
• CORBA (Common Object Request Broker Architecture)
• A middleware for WSN should facilitate
  development, maintenance, deployment and
  execution of sensing-based applications.

31
Agenda

• 1. Introduction of Sensor Network
• 1.1 Application Examples
• 1.2 Sensor Network Architecture
• 1.3 Hardware and Software
• 1.4 Sensor Network's Features
• 2. Middleware for Sensor Network.
• 2.1 Why do we need Middleware for SN?
• 2.2 Challenges in designing middleware.
• 2.3 Middleware approaches for SN.
• 2.4 Example Mires, A Message-Oriented Middleware

32
Challenges in designing middleware.

• Limited power and resources.
• Advance microelectronics technology allows tiny
  devices but limited in energy and resources, i.e.
  CPU and memory.
• Middleware should provide mechanisms for
  efficient processor and memory use while enabling
  lower-power communication and prolong sensor node
  lifetime. i.e. sleep mode, minimize number of
  transmission.
• Scalability, mobility, and dynamic network
  topology.
• As the application grows, device failure, moving
  obstacles, mobility, and interference, the
  network will change frequently.
• Middleware should maintain performance and robust
  operation while network changes dynamically.
  Also, It should support mechanisms for fault
  tolerance and sensor node self-configuration and
  self-maintenance.
• Heterogeneity.
• CPU-power, networking, memory and storage,
  operating systems.
• Middleware should able to interface various kinds
  of hardware, software and networks.
• Dynamic network organization.
• In most situation, Client/Server mode is
  impossible, no infrastructure is not available.
• Middleware should support Ad-hoc capability to
  discover resource and its location which affect
  the trade-offs among latency, reliability, and
  energy.

33
Challenges in designing middleware.

• Real-world integration.
• Most of the applications are dealing with
  real-time phenomena.
• Middleware should provide real-time services.
• Application knowledge.
• Developer would like to inject application
  knowledge to the network so as to map application
  communication requirements to it,
• Middleware design should balance between
  application specificity and middleware
  generality.
• Data aggregation.
• Network generate lots of redundant data,
  communications cost is much higher than
  computational cost. Sending a single bit can
  consume the same energy as executing 1000
  instructions
• Middleware should able to aggregate data to
  eliminate redundancy and minimize the number of
  transmissions to the sink.
• Security.
• Middleware efforts should concentrate on
  developing and integrating security in the
  initial phases of software design, hence
  achieving different security requirements such as
  authentication, integrity, freshness, and
  availability.

34
Agenda

• 1. Introduction of Sensor Network
• 1.1 Application Examples
• 1.2 Sensor Network Architecture
• 1.3 Hardware and Software
• 1.4 Sensor Network's Features
• 2. Middleware for Sensor Network.
• 2.1 Why do we need Middleware for SN?
• 2.2 Challenges in designing middleware.
• 2.3 Middleware approaches for SN.
• 2.4 Example Mires, A Message-Oriented Middleware

35
Middleware approaches for WSN.

• Virtual Machine (Cluster-Based)
• Because of its similarity to the virtual machine
  concept in traditional distributed systems in
  terms of providing application semantic
  transparency from the physical infrastructure.
• Pros
• Common abstraction.
• Sand-boxing.
• Cons
• High overhead
• Difficult to exploit heterogeneity.
• Example 1 - Maté
• Power-centric abstraction.
• Tied to TinyOS.
• Broken up into 24 byte-long instruction capsules,
  easy for distribution.
• Provide simple programming interface to sensor
  nodes. i.e. 6 instructions only for sense and
  send program.
• Communication in synchronous, less complex in
  programming.
• No support for message buffering / large storage.

36
Middleware approaches for WSN.

• Example 2, MagnetOS
• Power-aware, adaptive OS.
• The whole network appears as a single JVM.
• Standard Java programs are re-written by MAGNET
  as network components.
• Components may then be injected into the
  network using a power-optimized scheme.

37
Middleware approaches for WSN.

• Mobile Agents (Modular programming)
• Pros
• Only parts of the program need to be updated,
  propagate efficiently.
• Cons
• High overhead, Doesnt allow hardware
  heterogeneity.
• Example
• IMPALA
• As modular as possible, efficiency of updates and
  support dynamic applications.
• The nature of its code instruction doesn't allow
  hardware heterogeneity.
• Application Adaption with different profiles
  possible. (energy efficient)
• Use in the ZebraNet project (wildlife monitoring).

38
Middleware approaches for WSN.

• Database
• Pros
• Entire sensor network is abstracted as a virtual
  relational database. Ease to interoperate with
  existing systems.
• Cons
• Doesnt support real-time applications, and
  provides only approximate results.
• Example
• Cougar
• Represents all sensors and sensor data in a
  relational database.
• Control of sensors and extracting data occurs
  through special SQL-like queries.
• Allows the scheduling of ongoing queries that
  provide incremental results.
• Decentralized Implementation, message passing
  based on controlled flooding.
• SINA (System Information Networking Architecture)
• Based on a spreadsheet database, wherein network
  is a collection of data-sheets and cells are
  attributes.
• Attribute-based naming, e.g. typetemperature,
  locationN-E, temperature50.
• Queries again performed in an SQL-like language.
• Decentralized Implementation based on clustering.

39
Middleware approaches for WSN.

• Message Oriented
• Pros
• Use Publish-subscribe to support asynchronous
  communication, allowing a loose coupling between
  the sender and the receiver
• Cons
• Overhead
• Example
• MIRES
• Publish-subscribe
• Multi-Hop Routing
• Additional Service (e.g. data Aggregation)
• Sense advertise over P/S and route to Sink.

40
Agenda

• 1. Introduction of Sensor Network
• 1.1 Application Examples
• 1.2 Sensor Network Architecture
• 1.3 Hardware and Software
• 1.4 Sensor Network's Features
• 2. Middleware for Sensor Network.
• 2.1 Why do we need Middleware for SN?
• 2.2 Challenges in designing middleware.
• 2.3 Middleware approaches for SN.
• 2.4 Example Mires, A Message-Oriented Middleware

41
A Message-Oriented Middleware for Sensor Networks
- Mires

• In general, it facilitates the development of
  network-applications over the WSN and providing
  common application services.
• Problem Thousands of sensor nodes and redundant
  data. Low availability of resources and
  processing capacity of the sensor nodes.
• How does it help Message-oriented which
  aggregate data, Multi-Hop routing and greatly
  reduce the among of transmissions, save lots of
  energy.

42

• Traditional request/response approach is not
  suitable for event-driven communication model.
  Publish/subscribe approach is used to query and
  extract data from the network.
• In applications Use in habitat monitoring,
  object tracking, precision agriculture, building
  monitoring and military systems.

43
MIRES Architecture

• Publish/Subscribe service
• communication between middleware services.
• Advertising the topics available.
• Maintaining the list of topics subscribed by the
  node application
• Publishing messages.
• Routing
• Multi-hop routing to the Sink
• 3 types of notification events
• TopicArrival,
• event signals that the node application has
  submitted data collected from sensors.
• StateArrival
• Event signals that data received from the
  network.
• TopicSetupArrival
• the subscribe message broadcasted from the user
  application.

44
Publish/Subscribe Service

• PublishState interface define the command used by
  ServiceX to publish their processing results.
• Notifier interface defines 3 events
• MultiHopRouter-route to the sink
• BCast-Boardcast Setup info.

45
Topic advertisement sequence diagram
46
Topic subscription sequence diagram
47
Data publishing sequence diagram
48
An Environment-monitoring application example
Temperature, humidity, sound and luminosity
sensors
Cluster Head
Cluster
Sink
49
Graphical User Interface Example
50
Conclusions

• Sensor Network can be used in many applications,
  such as Military, Environmental and Healthetc.
• Its characteristics are tiny node, low power,
  limited resources, dynamic network topology and
  various scales of network deployment.
• Middleware is used to connect the network
  hardware, operating systems, network stacks, and
  applications in different approaches.
• For examples, Virtual Machine, Mobile Agent,
  Database and Message Oriented.

51

• Thanks You !
• Questions?

52
Discussion

• Security in Sensor Networks.
• Public/Private Key
• Key establishment beyond sensor network
  capabilities.
• Shared Key
• Simple solution, but single node may reveal the
  secret key.
• Scalability? ? each node stores n-1 keys (n(n-1)
  keys need to be established)
• Solution?
• Privacy Aspects in Sensor Networks.
• Sensor technology may be used for illegal
  surveillance.
• Providing awareness of the presence of sensor
  nodes?
• Solution?