iCAR : an Integrated Cellular and Ad-hoc Relaying System *

1
iCAR an Integrated Cellular and Ad-hoc Relaying
System

• Hongyi Wu
• Advisor Dr. Chunming Qiao
• LANDER, SUNY at Buffalo

This project is supported by NSF under the
contract ANIR-ITR 0082916 and Nokia.
2
Outline

• Motivations
• Introduction of iCAR
• ARS Placement
• Seed ARS
• Quality of Coverage
• iCAR Performance
• Theorems
• Analysis
• Simulations
• Signaling Protocols
• Future Work and Conclusion

3
Outline

• Motivations
• Introduction of iCAR
• ARS Placement
• Seed ARS
• Quality of Coverage
• iCAR Performance
• Theorems
• Analysis
• Simulations
• Signaling Protocols
• Future Work and Conclusion

4
What is a cellular system?

• The problem of scarce frequency resource
• Based on subdivision of geographical area
• One Base Transceiver Station (BTS) in each cell.
• Frequency is reused in cells far away.

5
Problems in Cellular Systems

• A MH can only access the channels in one cell
  (except soft-handoff).
• Unbalanced traffic among cells
• Variable locations of the Hot Spots (congested
  cells)
• Cell-splitting not flexible nor cost-effective
  enough
• Tremendous growth of wireless data/voice traffic
• Limited capacity

6
What is Mobile Ad hoc Network (MANET)?

• An autonomous system of mobile nodes connected by
  wireless links.
• The nodes are routers.
• The nodes are organized in a arbitrary graph.
• The nodes are free to move.

7
Objectives of Our Work

• Balance traffic among cells
• Decrease call blocking and dropping probability
• Increase systems capacity cost-effectively
• Support heterogeneous networks
• Provide service for shadow area
• Reduce mobile hosts (MH) transmission power
  and/or increase transmission rate

8
Outline

• Motivations
• Introduction of iCAR
• iCAR Placement
• Seed ARS
• Quality of Coverage
• iCAR Performance
• Theorems
• Analysis
• Simulations
• Signaling Protocols
• Future Work and Conclusion

9
Basic Idea Integration of Cellular and Ad-hoc
Relaying Technologies

• ARS Ad-hoc Relaying Stations
• Each ARS and MH has two interfaces (celluar and
  relay)

ARS
MH
10
One example of relaying

• MH X moving into congested Cell B is relayed to
  Cell A

x
A
B
A
B
x
(a)
(b)
11
An ARS differs from a BTS and a MH

• Compared to BTS
• Mobility
• Air interface
• Compared to MH
• Mobility
• Security,authentication,privacy
• Billing

12
Basic Operations

• Primary Relay a strategy that establishs a
  relaying route between a MH (in congested cell)
  to a nearby non-congested cell.
• Failed Hand-off
• Blocked new call
• MH switches over
• from C-interface
• to R-interface

A
B
x
13
Basic Operations (Contd)

• Secondary Relay
• Primary relay failed
• Not covered by ARS
• Reachable BTS is congested too
• Free the channel of an active call which can be
  relayed to a neighbor cell

x
A
B
y
(a)
x
A
B
y
(b)
14
Basic Operations (Contd)

• Cascaded Relay
• Cascade the above relays more multiple times if
  they are failed.

x
x
A
B
A
B
y
y
z
z
C
C
15
CI and NCI

• Congestion-Induced (CI) Relaying
• Reduce call blocking or dropping probability when
  congestion occures.
• Noncongestion-Induced (NCI) Relaying
• Pro-actively balance load
• Shadowing Area
• Uncovered Area
• Transmission Power

16
Outline

• Motivations
• Introduction of iCAR
• ARS Placement
• Seed ARS
• Quality of Coverage
• iCAR Performance
• Theorems
• Analysis
• Simulations
• Signaling Protocols
• Future Work and Conclusion

17
Full Coverage

• The maximum number of relay stations needed so as
  to ensure that a relaying route can be
  established between any BTS and an MH located any
  where in the cell

2 Km
1.5 Km
1 Km
200m
50
200
114
18
350m
66
38
500m
8
18
32
18
Seed Growing Approach

• With fewer ARSs, relaying can still be
  effective. Some can be seeds (placed at each pair
  of shared edges), and others can grow from them
  (placed nearby).

19
Number of Seed ARSs

• For a fix coverage area, the system with fewer
  UN-SHARED edges needs more seed ARSs.
• The max number is obtained by considering a
  circle area and count the number of shared edges.

Proposition For a n-cell system, the maximum
number of seed ARSs is
20
Quality of Coverage

• The quality of ARS coverage (Q) is defined to be
  the relay-able traffic in an iCAR system.
• The Q value depends on the traffic intensity, the
  cell size, the ARS size, the system topology,
  etc.
• The higher the Q value, the better the ARS
  placement
• The Q value is not always proportional to the ARS
  coverage.

21
Seed ARSs Placement
B

• Two approaches to place the seed ARS
• Edge (ARS No.1)
• S ARS ceverage
• TA, TB Traffic intensity of cell A and B.
• bA,bB Blocking probability of cell A and B.

B
B
A
2
2'
1
B
B
1'
B
Seed ARSs
3
3'

Half of S covers cell A, but only unblocked
part (1-bB) of them is relay-able
22
Seed ARSs Placement
B

• Vertex (ARS No.1')

B
B
A
2
2'
1
B
B
1'
B
Two third of S covers cell B. ..
One third of S covers cell A. Note that, the
Blocking probability is bB2 because the call
may Be relayed to two cells.
Seed ARSs
3
3'
23
Seed ARSs Edge v.s. Vertex

• Preliminary results
• Case1 when TBltTAlt50 Erlangs, QvertexltQEdger.
• Case2 when TA, TBgt50 Erlangs or TAltTB,
  QVertexgtQEdge.
• Case2 is out of normal operation range
• Rule of Thumb 1
• Place the seed ARS's at edges of a hot spot cell.

24
Seed ARS v.s. Grown ARS

• Preliminary Results
• Case1 seed (ARS 2). Assuming edge placement of
  seed)
• Case2 grow (ARS 2). The QoC value of the grown
  ARS is about 0.61S TA(1-bB).
• Rule of Thumb 2
• Try to place an ARS as a seed if it is possible.

25
Growing Direction

• When there are already sufficient seed ARSs,
• Additional ARS's can grow
• toward inside of a hot cell A (ARS No.3)
• toward outside of cell A (ARS No.3')
• Rule of Thumb 3
• Place an ARS in the cell with a higher traffic
  intensity.

26
Outline

• Motivations
• Introduction of iCAR
• ARS Placement
• Seed ARS
• Quality of Coverage
• iCAR Performance
• Theorems
• Analysis
• Simulations
• Signaling Protocols
• Future Work and Conclusion

27
Theorems

• Theorems1
• Assume that the total traffic in a n-cell system
  is T Erlangs, then the (system wide) call
  blocking probability is mininized when the
  traffic in each cell is T/n Erlangs.
• Why?
• Assume there are M channels in each cell, and the
  traffic intensity in cell i is Ti (
  ). According to Erlang B formula, the
  blocking probability in each cell is

28
Theorem (Contd)

• The average blocking probability of entire
  system is

In order to compute the minimum value of B, we
derive the partial differentiation,
Solve a group of equations, we can get the
critical points,
29
Theorems (Contd)

• Theorem2
• For a given total traffic in a system, and a
  fixed number of data channels, an idea iCAR
  system has a lower blocking probability than any
  conventional cellular system (including a
  perfectly load-balanced one).
• Why?
• An idea iCAR system can relay traffic from one
  cell to any other cells. So, it can be treated as
  a SUPER cell with nT traffic and nM channels. The
  blocking probability of the super cell is
• We can prove that it is lower than B(M,T).

30
Analysis based on multi-dimensional Markov chains

• Consider a system with only seed ARSs

31
Analysis (Contd)

• For primary relaying
• An approximate model (considering cell X in
  figure (b))
• To simplify the analysis, we assume that the
  blocking probability of the neighboring cells of
  X is fixed, i.e. the traffic relayed to cell Bs
  wont change their blocking probability. This
  assumption will be nullified in the accurate
  analysis model.

32
Analysis (Contd)

• For primary relaying
• An approximate model
• State diagram
• Final result

33
Analysis (Contd)

• For primary relaying

34
Analysis (Contd)

• An accurate model of primary relaying for a
  2-cell system.

35
Analysis (Contd)

• Secondary relaying
• An approximate model

36
Analysis (Contd)

• An accurate model

37
Simulations

• Simulation model
• GloMoSim
• 25 cells
• Cell A is a hot spot
• Location dependent traffic (ripple effect)
• 50 DCHs per cell
• 56 seed ARSs
• 25,600 MHs
• Call arrive rate is in poisson distribution
• Holding time is in exponential

38
Simulations (contd)

• Results
• Blocking rate
• Blocking rate can be reduced by primary relaying,
  but not much
• Secondary relaying reduces the call blocking rate
  further

39
Simulations (Contd)

• More results

Throughput
Call Dropping Rate
40
Outline

• Motivations
• Introduction of iCAR
• ARS Placement
• Seed ARS
• Quality of Coverage
• iCAR Performance
• Theorems
• Analysis
• Simulations
• Signaling Protocols
• Future Work and Conclusion

41
Signaling and routing protocols for QoS traffic

• Why do we need signaling and routing protocols?
• For iCAR to support real-time IP-based
  applications in wireless mobile environment, set
  up bandwidth guaranteed relaying path.
• Candidates of protocols for iCAR

42
Protocol 1 a PSC-assisted protocol

• Primary relaying

43
Protocol 1 (contd)

• Secondary relaying

44
Protocol 2 a link-state based protocol

• Primary relaying

45
Protocol 2 (Contd)

• Secondary relaying

46
Protocol 3 an aggressive route-searching protocol

• Primary relaying

47
Protocol 3 (contd)

• Secondary relaying

48
Performance Comparison

• Three protocols have their own advantages and
  disadvantages
• The PSC-assisted protocol will have the lowest
  signaling overhead in terms of the number of
  signaling messages. But in this protocol, PSC
  becomes the performance bottle neck and a signal
  point of failure.

49
Performance Comparison (Contd)

• The link-state based protocol is distributed. It
  requires the ARSs to flood the update messages.
  Also, the ARSs need large enough memory to
  maintain topology and bandwidth information, and
  high computation power to compute the relaying
  route.
• The aggressive route searching protocol does not
  maintain the relaying bandwidth information of
  other ARSs. It is an on-demand and the simplest
  distributed protocol. It requires fewest memory
  and computing power.

50
Simulation

• We evaluate the performance of the proposed
  signaling protocols in terms of request rejection
  rate and signaling overhead via simulations.
• Seven cells, 3060 ARSs and 1600 MHs were
  simulated in the model we discussed before.

51
Simulation Results

• Blocking rate

52
Simulation results (contd)

• Signaling Overhead

53
Outline

• Motivations
• Introduction of iCAR
• ARS Placement
• Seed ARS
• Quality of Coverage
• iCAR Performance
• Theorems
• Analysis
• Simulations
• Signaling Protocols
• Future Work and Conclusion

54
Future Work

• Mobility Tracking
• With the help of GPS, we can keep track of the
  position of MHs and ARSs, so that we can move the
  ARSs to the best positions.
• ARS Management/Moving
• With the movement of ARSs, issues such as route
  reestablishment, etc., need to be addressed.

55
Future Works (Contd)

• MAC layer design
• The iCAR system needs a novel MAC protocol to
  support relaying. The IEEE802.1X protocols may
  not be the optimized solutions for iCAR as the
  cellular infrastructure can help packet
  scheduling so as to avoid collisions.

56
Conclusion

• A purely cellular or purely Ad-hoc network will
  not be scalable, nor versatile enough.
• The integrated architecture can efficiently
  balance the traffic load dynamically, thus reduce
  the call blocking /hand-off dropping probability,
  and increase the effective capacity of a system.
• Other benefits include shadow coverage, fault
  tolerance and reduced transmission power and/or
  increased transmission rate.

57
Publications

• Integrated Cellular and Ad hoc Relaying (iCAR)
  System Pushing the Performance Limits of
  Conventional Wireless Networks, HAWAII
  INTERNATIONAL CONFERENCE ON SYSTEM SCIENCES,
  HICSS-35, January 7-10, 2002, Big Island, Hawaii.
• Overcoming The Limits Imposed By Cellular
  Boundaries With iCAR", in Asia-Pacific Optical
  and Wireless Communications, November 12-16,
  2001. Beijing, China.
• "An Integrated Cellular and Ad hoc Relaying
  System iCAR", in IEEE Journal on Selected Areas
  in Communication (JSAC) special issue on Mobility
  and Resource Management in Next Generation
  Wireless System, Oct., 2001.
• "Distributed Signaling and Routing Protocols in
  iCAR (integrated Cellular and Ad hoc Relaying
  System)", in the Fourth International Symposium
  on Wireless Personal Multimedia Communications
  (WPMC'01), Sept. 9-12, 2001. Aalborg, Denmark.

58
Publications (Contd)

• "Quality of Coverage A New Concept for Wireless
  Networks", in ACM SIGCOMM 2001 conference student
  poster session, August 27-31, 2001, Mandeville
  Auditorium, UC San Diego, CA
• "Performance Analysis Of iCAR (Integrated
  Cellular and Ad-hoc Relay System)", in IEEE
  International Conference on Communications
  (ICC'01), June 11-14, 2001. HELSINKI, FINLAND.
• "An New Generation Wireless System with
  Integrated Cellular and Mobile Relaying
  Technologies", in International Conference on
  Broadband Wireless Access Systems (WAS'2000),
  Dec. 4-6, 2000. San Francisco, CA. 
• "iCAR an Integrated Cellular and Ad-hoc Relay
  System", in IEEE International  Conference on
  Computer Communications and Networks
  (ICCCN'2000), Oct, 2000. Las Vegas, NV.
• "Load Balancing via Relay in Next Generation
  Wireless Systems" in IEEE Workshops on Mobile Ad
  Hoc Net Working and Computing (MobiHoc'2000), in
  conjunction with MobiCom'2000, Aug 7-11, Boston,
  MA. pp. 149-150.