PCS channel assignment and handover schemes

1
PCS channel assignment and handover schemes

• ???
• ??????????????
• wklai_at_cse.nsysu.edu.tw

2
????

• ???????
• System Aspects of Cellular Radio
• Channel Assignment Algorithms
• ???????
• ??

3
System Aspects of Cellular Radio

• There are two main components in mobile radio
  systemsgt the radio interfacegt a fixed network
• What makes public cellular radio complex?gt the
  control structure
• The number of users a network can support is
  fundamentally dependent on the Common Air
  Interface (CAI) over which users communicate.

4

• Other factsgt the amount of spectrum the
  regulators allocategt the size of the radio
  coverage area from a BSgt the amount of
  interference a particular radio link can
  tolerate.

5
System Planning in First- and Second-Generation
Systems

• Hand Offgt Signaling between the mobile, the
  BSs, and control centers
• Clustergt formed by cell(s)gt uses the entire
  allocated spectrum.
• Co-channel interferencegt using the same channel
  between two mobiles.gt is contained to acceptable
  limits by the distance between the cells.

6
System Planning in First- and Second-Generation
Systems

• If the cell sizes are decreased which causes a
  corresponding reduction in cluster size, the
  number of channels per unit area increases.
• The most effective way of increasing network
  capacity is to decrease the cell size, but the
  complexity of the network infrastructure
  increases.

7
Growth Scenarios

• In the start-up phase of a cellular network,
  capacity is not the problem.
• As the network matures, capacity becomes
  increasingly important.
• Cluster size is decreased while maintaining
  signal-to-interference ratios (SIRs) that ensure
  that link quality is acceptable.
• Sectorization generally results in an increase in
  SIR.
• Sectorization must be introduced with decreasing
  of cluster size.

8
Growth Scenarios

• The minimum acceptable SIR (SIRmin) is
  system-specific.
• By using discontinuous transmission(DTX),
  frequency hopping (FH) of the carrier and power
  control, such systems can allow lower SIRmin.

9
Microcells

• "Microcells" are used in CT-2, PCS and DECT.
• Conventional microcells are interconnected to
  mobile switching centers typically in a star
  configuration via standard transmission
  facilities, such as 1.5Mb/s(North-American TI
  Standard) or 2 Mb/s(European) links.
• Some microcells are essentially "remote radiation
  sites".

10
HandOver Issues in FDMA and TDMA

• HandOver(HO), or HandOff is the switching
  procedure when a MS changes its communication
  from one BS to an adjacent one when the received
  signal decreases below a system threshold.
• There are two type of HOSoft HO(SHO) and Hard
  HO(HHO).
• HHOgt break before make.gt communications of a
  MS with a BS are served before they are
  re-established with the new BS.

11
HandOver Issues in FDMA and TDMA

• SHOgt both the existing BS and the BS that will
  ultimately assume responsibility for the call
  communicate simultaneously with the MS.
• Preparing for an HOgt the decision when and
  where to HO must be made.gt both the handset and
  the network must switch.
• The decision algorithm typically uses
  measurements of received signal strength
  indication (RSSI) and bit error rate (BER) to
  detect the need to HO and must identify a free
  channel in a neighboring cell.

12
HandOver Issues in FDMA and TDMA

• Prioritized HO schemesgt some aim to minimize
  both the probability of forced termination of
  calls in progress due to HO failures, and the
  degradation in spectrum utilization.gt others aim
  at balancing or dissipating the teletraffic load
  across neighboring cells.

13
DCA in FDMA and TDMA Systems

• FCA Fixed Channel Assignment.
• Dynamic Channel Assignment (DCA) can in principle
  operate with FDMA or TDMA and with only modest
  enhancements to second-generation.
• The assignment of channels may be done by a
  system that adapts to both the traffic loading at
  the BS and the interference on the channels.

14
DCA in FDMA and TDMA Systems

• Traffic-adaptive systemgt channel borrowing in
  a traffic-adaptive system, when a cell is
  overloaded, it can use idle channel of its
  neighboring cell.gt Markov allocation assigns to
  each new call the first unused and
  non-interference channel in an order specific to
  the corresponding cell-site.

15
DCA in FDMA and TDMA Systems

• Interference-adaptive systemgt lends itself to
  distributed DCA algorithms that are able to
  self-organize.gt The MS controls the channel
  assignment of a call without the need to
  communicate with other BSs or with a central
  controller.
• The choice among acceptable channels is an
  important issue choosing the least-interfered
  channel provides most robust quality, while
  selection according a pre-defined order may lead
  to great capacity.

16
DCA in FDMA and TDMA Systems

• By specifying the SIR thresholds to be higher
  than the minimum required for good link quality,
  robustness against measurement and decision
  errors may be achieved.
• A simple algorithm used in transmitter power
  control is for each user to increase its transmit
  power when the SIR is inadequate and to decrease
  the SIR when it is more than adequate, and this
  succeeds whenever there is any set of power
  levels that allow all the stations to achieve an
  adequate SIR.

17
DCA in FDMA and TDMA Systems

• Power control greatly enhances performance, and
  the DCA process need not cause unacceptable
  channel reassignment or call-failure rates even
  in the presence of high user demand.
• The best way of integrating interference-adaptive
  power control with channel assignment is
  currently unknown.
• Important issues of DCAgt re-assignment may be
  necessary.gt Delay must be controlled.gt SIR
  measurement and channel-search techniques.gt
  TDMA-based DCA requires base-to-base synchrony
  for full capacity gain.

18
Code Division Multiple Access CDMA Systems

• Of paramount importance is the fact that CDMA
  uses single-cell clusters.
• No frequency planning is required in CDMA
  networks due to the on-cell clusters which all
  use the same carrier frequency.
• SHO is used in CDMA to avoid near-far problems at
  the cell-edge due to cell-membership ambiguity.
• Softer SHO (SSHO) is used between sectors of the
  same cell.

19
Channel Assignment Algorithms

• PCS ????
• PCS ????
• Wireless Resource
• Wired Resource
• Mobility Management Resource
• Cells for frequency reuse

20
(No Transcript)
21

• Channel ?????
• Hand-off channel assignment request
• Initial call channel assignment request
• Channel ????????
• Forced termination
• Blocked call

22

• ????
• ??? Mobile user??,?????????????????? , ????
  Channel assignment ?????????
• ????? NPS , RCS, FIFO, MBPS ????????????? ,
  ?????????????????????? ,???????
• 10 Pf Po

23
(No Transcript)
24

• ??????????
• NPS ( Non-Prioritized Scheme)
• RCS ( Reserved Channel Scheme)
• Queuing Priority Scheme (QPS)
• MBPS
• FIFO
• Genetic Algorithms

25
NPS ???

• Non-Prioritized Scheme(NPS)?Handoff channel
  assignment? Initial call channel assignment????,?
  cell ?? channel???,???Forced Termination ?
  Blocked Call????? channel ????? queuing ???

26
RCS ???

• Reserved Channel Scheme(RCS)???cell??channels????
  ,?????? Handoff ?Initial call??????? Handoff

27
Handoff area ??

• ?????cell ??????handoff area,??handoff area
  ???,??handoff ? channel ???????channel
  ????,???queuing ???
• ??? handoff area ?,??????? cell ?
  channel,???cell???????channel,?????handoff???,????
  ?cell???channel???handoff??????cell ?base
  station???queue?(?MS?? handoff area ?)???? cell
  ?channel???,????handoff????????Forced
  Termination????
• FIFO , MBPS ??? handoff area ??????

28
FIFO?MBPS ???

• ?????????? handoff area ???
• ??queue??First In First Out (FIFO),????MS
  ??handoff area ??(?????cell ?????),?????channel???
  priority,? Measure -Base Priority Scheme (MBPS)?

29
???????

• ????????
• ? blocked call ? forced termination ???????
• ??? ?? handoff ? channel ????? queueing ?? ,
  ???????? traffic load ???

30
?????????

• ?? blocked call ? forced termination ?????
• ???????? , ???? initial call ? handoff ???
  channel ?????? ??? ??????? , ?? channel
  ?????????????
• ??? handoff ? initial call??? channel ?????? ,
  ???? initial call queuing ???

31
(No Transcript)
32
A. Variables 5 tc the call holding time of an
MS
tdh the channel occupation time of a handoff
call tdo the channel occupation time of a new
call tm,i the residence time of an MS at i-th
cell ??? ??tc ? the mean call holding
time. ??? ??tm,i ? the mean MS residence time
in a cell 1/? the mean degradation interval
when the MS handoffs from a cell to a neighboring
cell ?o the new call arrival rate to a cell ?h
the handoff call arrival rate to a cell
33

• fc (tc ) the exponential density function of a
  call holding time tc
• fm (tm,i ) the exponential density function of
  tm,i
• fm(s) the Laplace transform of fm(tm,i)
• Pf the forced termination probability
• Po the new call blocking probability
• Pc the probability that a call is completed
• (neither blocked nor force-terminated).
• Pnc the probability that a call is not
  completed
• (either blocked or force-terminated)
• E(J) the expected number of handoffs when the
  call is forced terminated
• or successfully terminated
• (1- Po)E(J) the expected number of handoffs
  before a call terminated
• (either completes or is forced terminated or is
  initially blocked)

34

• Basic assumptions
• ?The incoming calls to an MS are a Poisson
  process.
• ?tc is the time duration between the beginning of
  a call and the completion of a call, the call
  holding time. It is assumed to be exponentially
  distributed with a density function fc(tc) and an
  average value ??? (i.e., and Etc 1/?).

35

• ?tm,i is the time duration that an MS stays in a
  cell i, the residence time of an MS at i-th cell.
  It is assumed to be exponentially distributed
  with a density function fm(tm,i) and an average
  value ??? (i.e., and Etm,i ???).

36
Basic Results

• PrK k is the probability of a K-handoff call
  ?????1 ? fm (?)2?fm(?)k?1 (1)
• 
  (2)
• Etdh Etdo 1 ? ?? ? ?)? (3)
• ?h ??1 Po??? ? ?Pf ) ?0 (4)

37

• Pnc 1 (1 Po)/(1 ?Pf /?) (5)

38

• Pc (the probability that a channel is
  available for initial access) (the probability
  that every handoff access is successful during
  the call holding time)
• 
  
• 
  (7)
• Pc ( 1 - Po ) ( 1 Pf ) Ek
• Po initial call channel ????????
• Pf handoff channel ????????
• Ek is the expected number of handoffs during
  the call holding time if the initial call is not
  blocked and is not forced terminated during
  handoffs

39

• A. We first derive the expected value of k.
• E k ? (Prk i) i (i 0 ?)
• From equation (1), Pr K k ?????1 ? fm (?
  )2? fm (? )k1
• Let A ?????1 ? fm (? )2 and B ? fm (? )
• E k A (B0 1) (B1 2) (B2 3)
• E k A /(1 B)2 (because 1 2X 3X2 4X3
  1 /(1 ? X )2 )
• ?????1 ? fm (? ) 2 1/1 ? fm (? )2
• ??? (9).
• Note that E(J) is equal to E(k) when Pf is zero
  and is less than E(k)
• when Pf is not zero.

40
B. Now we can begin to derive the optimal value
of Pc and the best proportion.

• y c?? ? ?) (10) from equation (3)
• Let the service rate for handling handoff calls
  be x, where x ? y. Then the service rates for
  handling initial calls are y x.
• We assume there are very few initial calls
  blocked and very few handoff calls forced
  terminated because of time-outs so they are
  negligible

41
Po 1 (y x)/?o, (y x) lt ?o (if (y x) ?
?o, then Po 0) Pf 1 (x /?h ), x lt ?h (if x
? ?h, then Pf 0) From equation (4), ?h ??1
Po??? ? ?Pf )?o We have Pf 1 (x/?h) 1
x(? ? ?Pf)/?o?(y ? x)/?o gt Pf ?(y ? x)
? x?/?y (11) From equation (8), Pc' (1 ?
Po)(1 Pf ) Ek Let Ek w Pc (1 ? Po) (1
Pf ) Ek (y x)/?o 1 ? ?(y ? x) ?
x?/?yw (? ? ?)wxw(y ? x) / (?y)w?o
42
Let C (?y)w?o and D (? ? ?)w Pc (D/C )xw(y
? x) (C, D, y, and w are constants) When xw(y ?
x) has the extreme value, the Pc has the extreme
value. We differentiate the xw(y ? x) and get the
following equation. dxw(y ? x)/dx wxw-1(y
x) xw (-1) Thus, when wxw-1(y x) xw (-1)
0, we have the extreme value. gt x (w/(w
1))y (12) If we differentiate xw(y ? x) twice, we
have the following equation. dwxw1(y x) xw
(-1) /dx w(w ? 1)xw2 (y ? x) wxw1( -1 )
wxw1 wwy/(w 1)w2(-y) lt 0 (x w y /(w
1))
43

• The approximation to the call completion
  probability, Pc, has the nice property of easier
  computation and manipulation.
• When the proportion of the handoff calls and the
  initial calls is w 1 (x (w/(w 1)) y), the
  alternative call completion probability, Pc, has
  the maximum value.

44

• Pc c?/ ?o, where Ek w ???. (13)
• The value of the approximation to the call
  completion probability is proportional to the
  number of channels on each cell and inversely
  proportional to the mean call holding time and
  the new call arrival rate to a cell. Its value is
  not dependent on the value of the mean MS
  residence time

45
C.

• When x (w/(w 1))y, the values of the call
  completion probability and the alternative call
  completion probability are equal.

46
If we substitute equations (9) and (10) into
equation (12), we have the following equation. x
(w/(w 1))y (???)c?? ? ?)/(???) 1 c?
(14) Then we substitute equations (10) and (14)
into equation (11) and get the following
equation. Pf ?(y ? x) ? x?/?y (c?? ?
c??)/?y 0 If Pf 0, Pc Pc 1 ? P0 (15).
47

• When Pf is not zero but close to zero, the
  difference between the call completion
  probability and the approximation to the call
  completion probability is very close to zero.

48

• Pc
• ? (1 Po)Pr(K 0) (1 Po) (1 Pf)Pr(K 1)
• (1 Po) (1 Pf)2Pr(K 2) (1 P0)
  (1 Pf)w
• (1 Pr(K 0) Pr(K 1) Pr(K w 1))
• ? Pc Pc ? (1 Po)Pr(K 0) (1 Po) (1
  Pf)Pr(K 1) (1 Po) (1 Pf)2Pr(K 2)
• (1 Po)(1 Pf)w1Pr(K w 1) (1 Po)
• (1 Pf)w(Pr(K 0) Pr(K 1) Pr(K w
  1))
• (1 Po)Pr(K 0)(1 (1 Pf)w) (1 Po)Pr(K
  1)
• (1 Pf)(1 (1 Pf)w1) (1 Po)Pr(K
  w 1)
• (1 Pf) w 1 (1 (1 Pf))
• ?(1 Po)(1 (1 Pf)w)

49

• If ? gtgt ?, then w 0. The difference between the
  call completion probability and the approximation
  to the call completion probability is equal to
  zero
• If ? is not far larger ?, the difference is equal
  to zero only if the value of the Pf is equal to
  zero

50

• In reality, the value of w is very close to 0 in
  most cases. For example, when the mean call
  holding time is 3 minutes and the mean portable
  residence time is 30 minutes, the value of w is
  1/10 and when the mean call holding time is 3
  minutes and the mean portable residence time is
  10 minutes, the value of w is 3/10.

51

• ????????????????????
• The queue size must be large enough
• The time-out for a handoff call can not be
  adjusted
• The time-out for an initial call can be adjusted
  within tolerable ranges

52

• ?????????????
• Single queue or Dual queues
• FIFO or Priority queue
• ? initial call time_out value ? Statistical
  Multiplexing ???????

53

• SFTT ???
• ???? ( Single queue )
• channel ??????????? ( FIFO )
• ????????( Time out )
• ??? initial call ??( Time out )??? , ???? channel
  ?????????????

54
SFTT ???
55
SFTT ???
56

• SPTT ???
• ???? ( Single queue )
• ????????( Priority )?????
• ????????( Time out )
• ??? initial call ??( Time out )??? , ???? channel
  ?????????????

57
SPTT ???
58
SPTT ???
59

• DFTS ???
• ????( Dual queue )
• channel ??????????? ( FIFO )
• ????????( Time out )
• ??????(Statistical Multiplexing)?????????

60
DFTS ???
61
DFTS ???
62

• DPTS ???
• ????( Dual queue )
• ????????( Priority )?????
• ????????( Time out )
• ??????(Statistical Multiplexing)?????????

63
DPTS ???
64
DPTS ???
65
Simulation Results
66
(a) The forced termination probability.
(b) The new call blocking probability.
(C)The call incompletion probability.
Figure 10. Performance of six schemes. The mean
MS residence time is 30 minutes. The mean call
holding time is 3 minutes. The mean degradation
interval is 18 seconds. The number of channels in
each cell is 50
67
Figure 11. The SFTT scheme with different 1/ ?
under heavy traffic.
68
Figure 12. The SFTT scheme with different 1/?
under light load
69
??

• (1) Giving priority to handoff calls over initial
  calls would not yield better call completion
  probabilities in general
• (2) The proportions of handoff calls and initial
  calls will influence the call completion
  probabilities

70

• (3) The implementation of the priority scheme has
  the effect of decreasing the call incompletion
  probabilities. However, it might also have the
  negative effect of increasing the forced
  termination probabilities.

71

• (4) The implementation of the statistical TDM has
  the effect of decreasing the call incompletion
  probabilities when the average new call arrival
  rates are high. It is because when there are many
  new calls, the ratio of initial calls and handoff
  calls served can be tuned with the statistical
  multiplexing.

72

• (5) The average values of time-outs for initial
  calls and handoff calls are another main factor
  for the proportions between handoff calls and
  initial calls. The values of time-out for initial
  calls can be adjusted to some degree and still
  within tolerable ranges, especially when portable
  data communications becomes popular.

73

• (6) If the new call arrival rates are not very
  high, longer average time-outs for initial calls
  have lower call incompletion probabilities. When
  the mean arrival rate is high, 80 Erlangs in our
  simulation, the longer average time-outs in
  contrast have higher call incompletion
  probabilities. Because if we have longer average
  time-outs for initial calls when the mean arrival
  rate is high, the proportions of initial calls
  being served are too many.

74
References

• System Aspects of Cellular Radio, IEEE
  Communications Magazine, January 1995 by Raymond
  Steele, James Whitehead, and W. C. Wong
• Introduction to Mobile Network Management by
  Yi-Bing Lin, National Chiao Tung University
  Series in Telecommunication, ?????, 1997.
• Channel assignment for initial and handoff calls
  to improve the call completion probability,Wei
  Kuang Lai, Yu-Jyr Jin, Hsin Wei Chen, and Chieh
  Ying Pan, IEEE Transactions on Vehicular
  Technology, vol. 52, no. 4, pp. 876 - 890, July
  2003.