location based spatial query processing in wireless broadcast environments

LOCATION BASED SPATIAL QUERY PROCESSING IN WIRELESS BROADCAST ENVIRONMENTS

• Geospatial data or geographic information.

• Identifies the geographic location of features and

boundaries on Earth(oceans, rivers, hospitals, restaurants etc).

• Usually stored as coordinates and topology.

• Spatial data is often accessed, manipulated or analyzed through Geographic Information Systems (GIS).

What is spatial data?

• A spatial query is a special type of database query supported by geodatabases.

• It’s a query which selects features based on their location or geographic relationship to others.

• Spatial query processing is the process of selecting features based on location or spatial relationship.

• For example, “Send me the status of a particular route”.

What are spatial queries?

• Location-based spatial queries (LBSQs) refer to spatial queries whose answers rely on the location of the inquirer.

• They represent a set of spatial queries that retrieve information based on mobile users’ current locations.

• Efficient processing of LBSQs is of critical importance with the ever-increasing deployment and use of mobile technologies.

• For example, "Find the top-three nearest hospitals of some location x”.

What are location based spatial queries?

• ON DEMAND DATA ACCESS MODEL

• WIRELESS BROADCAST MODEL

• P2P DATA SHARING MODEL

Approaches of mobile data access for spatial query processing

• It’s a simple client-server model.

• In the simplest approach, a user establishes a point-to-point communication with the server so that his queries can be answered on demand.

• Here server queues up the query requests issued by the clients and processes them accordingly. The results of query processing are returned to the mobile user through the same point-to-point link.

ON DEMAND DATA ACCESS MODEL (Traditional centralized server model)

Queue

Point to point link

Client

c1c2c3c4

ON DEMAND DATA ACCESS MODEL

• First, it may not scale to very large user populations.

• Second, to communicate with the server, a client must most likely use a fee-based cellular-type network to achieve a reasonable operating range.

• Third, users must reveal their current location and send it to the server, which may be undesirable for privacy reasons.

• Fourth, it is subjected to single point failure of server which disrupts the entire system.

Disadvantages of ON DEMAND DATA ACCESS MODEL

• In the Wireless broadcast model, the server repeatedly broadcasts all the information in wireless channels, and the clients are responsable for filtering the information. An example of such a system is the Microsoft DirectBand Network.

• To facilitate information retrieval on wireless broadcast channels, the server usually transmits an index structure, along with data objects. A well-known broadcast index structure is the (1, m) indexing allocation method.

WIRELESS BROADCAST MODEL

The general access protocol for retrieving data on a wireless broadcast channel involves three main steps

• The initial probe

• Index search

• Data retrieval

• Its a more advanced solution.

• It can support an almost-unlimited number of mobile hosts (MHs) over a large geographical area with a single transmitter.

• With the broadcast model, MHs do not submit queries, Instead, they tune in to the broadcast channel for information that they desire. Hence, the user’s location is not revealed and his privacy is retained.

• The main advantage of the broadcast model over the on-demand model is that it is a scalable approach.

Advantages of WIRELESS BROADCAST MODEL

Limitations of WIRELESS BROADCAST MODEL

1. The broadcast model has large latency, as clients have to wait for the information that they need in a broadcasting cycle. Furthermore if a client misses the packets that it needs, it has to wait for the next broadcast cycle.

2. Nearly all the existing spatial access methods are designed for databases with random access disks. These existing techniques cannot be used effectively in a wireless broadcast environment, where only sequential data access is supported.

3. Since there is significant delay in answering the spatial queries, the answers provided become invalid especially in case of mobile nodes.

4. Queries can only be fulfilled after all the required on-air data arrives.

• The main limitation of preceding on air KNN query lies in its sequential data access: the access latency becomes longer as the number of data items increases. If we can provide (approximate) answers to spatial queries before the arrival of related data packets, we will overcome the limitation of the broadcast model.

•The fundamental idea behind our methodology is to leverage the cached results from prior spatial queries at reachable MHs for answering future queries at the local host. This is known as P2P cooperative caching with result sharing.

•A novel component in our methodology is a verification algorithm that verifies whether a data item from neighboring peers is part of the solution set to a spatial query.

Communication range of q

P1

P1| P2

P2|

O2q

O4

O31NN candidate

1NN candidate

o1

P2P cooperative caching with result sharing.

Mobile host transmission range

Wireless Broadcast Channel

Peer-to-Peer Channel

Mobile Host

Data station

System environment

Spatial Database

POINT OF INTEREST

A POI, is a specific point location that someone may find useful or interesting. Ex:Hospital,resturant etc.

MINIMUM BOUNDED RECTANGLE

A rectangle, oriented to the x and y axes, which bounds a geographic feature or a geographic data set. It is specified by two coordinates: xmin,ymin and xmax,ymax.

• Any MH ‘p’ exclusively belongs to an immediatly enclosing MBR at any instant of time.

MBR

(xmax,ymax)

(xmin,ymin)p

VERIFIED REGION

Since memory space is scarce in mobile devices, we assume that each MH p caches a set of POIs in an MBR related to its current location.

• Since the POIs located inside the MBR were obtained from the wireless information server, we define the area bounded by the MBR as verified region p.V R with regard to p’s location.

Sharing based nearest neighbor verification

• When an MH q executes SBNN, it first broadcasts a request to all its single-hop peers for their cached spatial data.

•Each peer that receives the request returns the verified region MBR and the cached POIs to q.

•Then, q combines the verified regions of all the replying peers, each bounded by its MBR, into a merged verified region MVR . The merging process is carried out by the MapOverlay algorithm.

•The core of SBNN is the NN verification (NNV) method, whose objective is to verify whether a POI oi obtained from peers is a valid (that is, the top-k) NN of the MH q.

Sharing based nearest neighbor verification

• Let P denote the data collected by q from j peers p1,p2…..pj. Consequently, the merged verified region MVR can be represented as MVR=p1.VR U p2.VR U …. U pj.VR.(MAPOVERLAY algorithm)

• Suppose that the boundary of MVR consists of k edges, E={e1,e2,…..ek}, and there are l POIs, O={o1,o2…..ol}, inside the MVR. Let es E be the edge that has the shortest distance to q. An example is given in fig where k=10, and e1 has the shortest distance to q.

SBNN

At neighboring nodes

1. Let N = { m1,m2,….. ….. mn } be the set of mobile nodes

2. If a node mi receives a broadcast request from a query mobile host ‘q’

3. If q.POI = mi.POI4. mi sends a response message to q consisting of

1. Its spatial data as a set of POIs’ present in its cache and

2.MBR information.

At ‘q’ : Algorithm: NNV (q, H, k)

• Euclidian distance between two points p1(x1,y1) and p2(x2,y2) is given by ||p1,p2||= (x2-x1)2+(y2-y1)2

Enhancement to the proposed solutionPROFILE BASED SBNN

• Need for profile based system is to eliminate irrelevant data processing.

• If we are able to categorize the spatial data and even the device users based on some criteria, it would give rise to a state where ‘q’ receives only those spatial data items that are tailored its comfort level.

• This thesis give rise to an innovative concept called as “PROFILE” of a user.

• Hence we put forth a new strategy for addressing the categorization of users by making the ordinary SBNN system to be equipped with what is called as “Profile based filtering”.

• It makes the query mobile host to receive and process only the relevant data tailored to its profile thus gaining advantage over ordinary SBNN.

At neighboring nodes,

1. Let N = { m1,m2,….. ….. mn } be the set of mobile nodes each with a profile from the set PR = {pr1,pr2,pr3}

2. If a node mi with profile prmi receives a broadcast request from a query mobile host ‘q’ with profile prq

3. If q.POI = mi.POI and prmi = prq

4. mi sends a response message to q consisting of 1. Its spatial data as a set of POIs’ present in its

cache and2. MBR information.

At ‘q’ : Algorithm: NNV (q, H, k)

Sharing based nearest neighbor verification

• The NNV method uses a heap H to maintain the entries of verified and unverified POIs discovered so far . Initially, H is empty. The NNV method inserts POIs to H as it verifies objects from MHs in the vicinity of q.

• The heap H maintains the POIs in ascending order in terms of their Euclidean distances to q. Unverified objects are kept in H only if the number of verified objects is lower than what was requested by the query.

• If k elements in H are all verified by NNV, the kNN query is fulfilled. There will be cases when the NNV method cannot fulfill a kNN query. Hence, a set that contains unverified elements is returned. If the response time is critical, a user may agree to accept a kNN data set with unverified elements, where the objects are not guaranteed to be the top kNNs, otherwise he has to switch into broadcast channel to complete the KNN query.

Node ID TransmissionRange

Cache Capacity

No. of Nodes No. of POIs % Increase in No. of nodes & POIs

6

150 3 11 33 -

175 3 20 60 81.81

195 3 28 84 40

215 3 34 102 21.42

PERFORMANCE OF SBNN

Results

PERFORMANCE OF SBNN

Node ID TransmissionRange

Cache Capacity

No. of Nodes

No. of POIs % Increase in No. of nodes & POIs

6

150 2 11 22 -

150 3 11 33 32

150 4 11 44 43

SBNN vs. PFSBNN

Node ID TransmissionRange

Cache Capacity

SBNN PFBNN

% Decrease

in nodes & POIs

No. of Nodes No. of POIs No. of Nodes No. of POIs

6 150 3 16 48 10 30 37.5

27 150 3 13 39 12 36 7.69

15 150 3 15 45 8 24 46.6

31 150 3 13 39 11 33 15.38

34 150 3 16 48 12 36 25

258 150 3 10 30 7 21 30

Node id->6

Graphs

Node id->6

Node id->6

conclusion

• With this novel SBNN algorithm, The delay in answering the KNN query is significantly reduced as it doesn’t need to filter all the information required to satisfy the query.

Future enhancement

• An efficient caching strategy and cache replacement strategy for distributed storage of spatial data can further more increase the efficiency of SBNN algorithm.