• 沒有找到結果。

Comprehensive Survey Systems: Channel Assignment Schemes Cellu Mobile Telecommunication

N/A
N/A
Protected

Academic year: 2022

Share "Comprehensive Survey Systems: Channel Assignment Schemes Cellu Mobile Telecommunication"

Copied!
22
0
0

加載中.... (立即查看全文)

全文

(1)

A B S T R A C T

This article provides a detailed discussion of wireless resource and channel allocation schemes. The authors provide a survey of a large number of published papers in the area of fixed, dynamic, and hybrid allocation schemes and compare their trade-offs in terms of

complexity and performance. We also investigate these channel allocation schemes based on other factors such as distributed/centralized control and adaptability to traffic conditions. Moreover, we provide a detailed discussion on

reuse partitioning schemes, the effect of handoffs, and prioritization schemes. Finally, we discuss other important issues in resource allocation such as overlay cells, frequency planning, and power control.

Channel Assignment Schemes for Cellu Mobile Telecommunication Systems:

A Comprehensive Survey

1.

KATZELA

A N D

M. N A G H S H I N E H

development of handheldFireless terminals have facilitated the rapid growth of wireless communications and mobile com- puting. Taking ergonomic and economic factors into account, and considering the new trend in th e telecommunications industry to provide ubiquitous information access, the popula- tion of mobile users will continue to grow at a tremendous rate. Another important developing phenomenon is the shift of many applications to multimedia platforms in order to pre- sent information more effectively.

The tremendous growth of the wirelessimobile user popu- lation, coupled with the bandwidth requirements of multime- dia applications, requires efficient reuse of the scarce radio spectrum allocated to wirelessimobile communications. Effi- cient use of radio spectrum is also important from a cost-of- service point of view, where th e number of base stations required to service a given geographical area is an important factor. A reduction in the number of base stations, and hence in the cost of service, can be achieved by more efficient reuse of the radio spectrum. The basic prohibiting factor in radio spectrum reuse is interference caused by the environment or other mobiles. Interference can be reduced by deploying effi- cient radio subsystems and by making use of channel assign- ment techniques.

In the radio and transmission subsystems, techniques such as deployment of time and space diversity systems, use of low- noise filters and efficient equalizers, and deployment of effi- cient modulation schemes can be used to suppress interference and to extract the desired signal. However, co-channel inter- ference caused by frequency reuse is the most restraining fac- tor on the overall system capacity in the wireless networks, and the main idea behind channel assignment algorithms is to make use of radio propagation path loss [l, 21 characteristics in order to minimize the carrier-to-interference ratio (CIR) and hence increase the radio spectrum reuse efficiency.

The focus of this article is to provide an overview of differ- e n t channel assignment algorithms and compare them in terms of performance, flexibility, and complexity. We first start by giving an overview of the channel assignment problem

in a cellular environment and discuss the general idea behind major channel allocation schemes. Then we proceed to discuss different channel allocation schemes within each category.

Channel Allocation Schemes

What Is Channel AIlocation?

A given radio spectrum (or bandwidth) can be divided into a set of disjoint or noninterfering radio channels. All such chan- nels can be used simultaneously while maintaining an accept- able received radio signal.' In order to divide a given radio spectrum into such channels many techniques such as fre- quency division (FD), time division (TD), or code division (CD) can be used. In FD, the spectrum is divided into disjoint frequency bands, whereas in T D the channel separation is achieved by dividing the usage of the channel into disjoint time periods called time slots. In CD, the channel separation is achieved by using different modulation codes. Furthermore, more elaborate techniques can be designed to divide a radio spectrum into a set of disjoint channels based on combining the above techniques. For example, a combination of T D and FD can be used by dividing each frequency band of an FD scheme into time slots. The major driving factor in determin- ing the number of channels with certain quality that can be used for a given wireless spectrum is the level of received sig- nal quality that can be achieved in each channel.

Let S i ( k ) be denoted as the set (i) of wireless terminals that communicate with each other using the same channel k . By taking advantage of physical characteristics of the radio environment, the same channel k can be reused simultaneous- ly by another set j if the members of sets i and j are spaced sufficiently apart. All such sets which use the same channel are referred to as co-channel sets or simply co-channels. The minimum distance at which co-channels can be reused with

In practice, each channel can generate some interference in the adjacent channels. However, the effect of such interference can be reduced by ade- quate adjacent channel separation.

10 1070-9916/96/$05.00 0 1996 IEEE IEEE Personal Communications June 1996

(2)

p 3

I

I D C A strategies have been intro-

I

acceptable interference is called the "co-channel reuse distance '' (5.

This is possible because clue to propagation path loss in the radio

\ duced.

I In DCA, all channels are placed d3 I in a ~ o o l and are assigned to new d.2 I

environment, t h e average power received from a transmitter at dis- tance d is proportional t o f ' ~ d - "

where

a

is a number in the range of 3-5 depending o n t he physical environment, and PT is the average transmitter power. Fo r example, for an indoor environment with a

= 3.5, the average power at a dis- tance 2d is about 9 percent of the average power received at distance d. Thus, by adjusting the transmit-

\ \ I <#\

.

,

Desired signal

/'

'. __

,,.' '\ PA

call; as n e e d e d such t h a t t h e CIRmi, criterion is satisfied. At the cost of higher complexity, D C A schemes provide flexibility and traf- fic adaptability. However, D C A strategies a r e less efficient than FCA under high load conditions.

T o overcome this drawback, HCA techniques were designed by com- bining FCA and DCA schemes.

.'ds

Interfering signals

Channel assignment schemes

- - - - -_ . - can be implemented in many dif-

ter power level and/or the distance between co-channels, a channel can be reused by a number of co-chan-

nels if the CIR in each co-channel is above the required value CIR,,,. Here the carrier (C) represents the received signal power in a channel, and the interference (I) represents the sum of received signal power:, of all co-channels.

As an example, consider Fig. 1 where a wireless station labeled R is at distance dt from a transmitter station labeled T using a narrowband radio channel. W e refer to the radio channel used by T t o communicate t o R as the reference channel. In this figure, we have also shown five other stations labeled 3 , 2, ..., 5 , which use the same channel as the refer- ence channel t o communicate with some o t h e r stations.

Denoting the transmitted power of station i by P, and the dis- tance of station i from R by d,, the average CIR at the refer- ence station R is given by:

1

P,d;"

CIR = (1) 5

x e d t - " +No

where No represents the environmental noise. To achieve a certain level of CIR at the reference station R , different meth- ods can be used. For example, the distance between stations 1, 2,

...,

5 using the co-channel and the reference station R can be increased to reduce the co-channel interference level.

Many channel allocation schemes are based on this idea of physical separation. Another solution to reduce the CIR at R is to reduce the interfering powers transmitted from five inter- fering stations and/or to increase the desired signal's power level Pr. This is the idea behind power control schemes. These two methods present t h e underlying concept fo r channel assignment algorithms in cellular systems. Each of these algo- rithms uses a different metlhod to achieve a CIR,,, at each mobile terminal by separating co-channels and/or by adjusting the transmitter power.

Different Channel Allocation Schemes

Channel allocation schemes can be divided into a number of different categories depending on the comparison basis. For example, when channel assignment algorithms are compared based on the manner in which co-channels are separated, they can be divided into fixed ch,annel allocation (FCA), dynamic channel allocation (DCA), and hybrid channel allocation (HCA).

In FCA schemes, the area is partitioned into a number of cells, and a number of channels are assigned to each cell according to some reuse pattern, depending on the desired signal quality. FCA schemes are very simple, however, they do not adapt to changing traffic conditions and user distribution.

In order t o overcome these deficiencies of F C A schemes,

ferent ways. For example, a chan- nel can be assigned to a radio cell based on the coverage area of the radio cell and its adjacent cells such that the CIR,,, is main- tained with high probability in all radio cells. Channels could be also assigned by taking the local CJR measurements of the mobile's and base station's receiver into account. That is, instead of allocating a channel blindly t o a cell based on worst-case conditions (such as letting co-channels be located at t h e closest boundary), a channel can be allocated t o a mobile based on its local CIR measurements [3, 41.

Channel assignment schemes can be implemented in cen- tralized or distributed fashion. In the centralized schemes the channel is assigned by a central controller, whereas in dis- tributed schemes a channel is selected either by the local base station of the cell from which the call is initiated or selected autonomously by the mobile. In a system with cell-based con- trol, each base station keeps information about the current available channels in its vicinity. Here the channel availability information is updated by exchange of status information between base stations. Finally, in autonomously organized dis- tributed schemes, the mobile chooses a channel based on its local CIR measurements without the involvement of a central call assignment entity. Obviously, this scheme has a much lower complexity at the cost of lower efficiency. It is impor- tant to note that channel assignment based on local assign- ment can be done for both FCA and DCA schemes.

Fixed Channel Allocation

n the FCA strategy a set of nominal channels is permanent- ly allocated to each cell for its exclusive use. Here a defi- nite relationship is assumed between each channel and each cell, in accordance to co-channel reuse constraints [5-121.

The total number of available channels in the system C is divided into sets, and the minimum number of channel sets N required to serve the entire coverage area is related t o the reuse distance s as follows [6, 321:

N = (1/3)02, for hexagonal cells ( 2 ) Her e (r is defined as DIR,, where R, is the radius of the cell and D is the physical distance between the two cell cen- ters [ 5 ] . N can assume only the integer values 3, 4, 7, 9, ... as generally presented by the series, ( i

+

j ) 2 - ij, with i and j

being integers [5, 71. Figures 2a and 2b give the allocation of channel sets to cells for N = 3 (o = 3) and N = 7 (0 = 4.45), respectively.

In the simple FCA strategy, the same number of nominal channels is allocated to each cell. This uniform channel distri- bution is efficient if the traffic distribution of the system is also uniform. In that case, the overall average blocking proba-

IEEE Personal Communications June 1996 1 1

~ ~ ~-

(3)

. .

I I

done in a scheduled or predictive manner, with changes in tively.

I

' traffic known in advance or based on measurements, respec-

Channel Borrowing Schemes

I

In a channel borrowing scheme, an acceptor cell that has used all its nominal channels can borrow free channels from its neighboring cells (donors) to accommodate new calls. A chan- nel can be borrowed by a cell if the borrowed channel does not interfere with existing calls. When a channel is borrowed, several other cells are prohibited from using it. This is called channel locking. The number of such cells depends on the cell layout and the type of initial allocation of channels to cells.

bility of the mobile system is the same as the call blocking For example, for a hexagonal planar layout with reuse dis- probability in a cell, Because traffic in cellular systems can be tance of one cell ( 0 = 3), a borrowed channel is locked in nonuniform with temporal and spatial fluctuations, a uniform three additional neighboring cells, as is shown in Fig. 3, while allocation of channels to cells may result in high blocking in for a one-dimensional layout or a hexagonal planar grid layout some cells, while others might have a sizeable number of with two-cell reuse distance, it is locked in two additional spare channels. This could result in poor channel utilization. neighboring cells.

It is therefore appropriate to tailor the number of channels in In contrast to static borrowing, channel borrowing strate- a cell to match the load in it by nonunlfom channel allocation gies deal with short-term allocation of borrowed channels to 113, 141 or static bowowing [15, 161. cells; once a call is com pl et ed, t h e borrowed channel is In nonuniform channel allocation the number of nominal returned to its nominal cell. The proposed channel borrowing channels allocated to each cell depends on the expected traf- schemes differ in the way a free channel is selected from a fic profile in that cell. Thus, heavily loaded cells are assigned donor cell to be borrowed by an acceptor cell.

more channels than lightly loaded ones. In [13] an algorithm, The channel borrowing schemes can be divided into simple namely nonuniform compact pattem allocation, is proposed for and hybnd. In simple channel borrowing schemes, any nominal allocating channels to cells according to the channel in a cell can be borrowed by a neigh- traffic distribution in each of them. Th e - - - - - boring cell for temporary use. In hybrid proposed technique attempts to allocate

1 -'.

[ channel borrowing strategies, t he set of channels t o cells in such a way t h a t t h e channels assigned t o each cell is divided average blocking probability in the entire into two subsets, A (standard or local chan- system is minimized. Let there be N cells nels) and B (nonstandard or borrowable and M channels in the system. The alloca- channels). Subset A is for use only in the tion of a channel to the set of co-channel nominally assigned cell, while subset B is cells forms a pattern which is referred to as allowed t o be l ent t o neighboring cells.

the allocation pattern 1131. In addition, the I I Table 1 summarizes the channel borrowing compact allocation pattern of a channel is schemes proposed in the literature. In the defined as the pattern with minimum aver- next two subsections we discuss the simple loads in each of the N cells and the possible

compact pattern allocations for the M chan- nels, the nonuniform compact pattem alloca-

tion algorithm at t empt s t o find t h e compatible compact patterns that minimize the average blocking probability in the entire system as nominal channels are assigned one at a time.

A similar technique for nonuniform channel allocation is also employed in the algorithms proposed in 1141.

Simulation results in [13] show that the blocking probabili- ty using nonuniform compact pattern allocation is always lower than the blocking probability of uniform channel alloca- tion. It is interesting to note that the reduction of blocking probability is almost uniformly 4 percent for the range of traf- fic shown in [13].2 Also for the same blocking probability, the system can carry, on the average, 10 percent (maximum 22 percent) more traffic with the use of the nonuniform pattern allocation [13].

In the static borrowing schemes proposed in [15, 161, unused channels from lightly loaded cells ar e reassigned to heavily loaded ones at distances 2 the minimum reuse dirtance 0. Although in static borrowing schemes channels are permanently assigned to cells, the number of nominal channels assigned in each cell may be reassigned periodical- ly according to spatial inequities in the load. This can be I

. - - - -

Figure 2. a ) N = 3; b) N = 7

! I

'@

Lockfn ,

age distance between cells. Given the traffic I - - - -

1

and hybrid borrowing schemes in detail.

Figure 3. Channel locking.

Simple Channel Borrowing Schemes - In the simple borrowing (SB) strategy [15-201, a nominal channel set is assigned t o a cell, as in t he FCA case. After all nominal channels are used, an available chan- nel from a neighboring cell is borrowed. To be available for borrowing, the channel must not interfere with existing calls.

Although channel borrowing can reduce call blocking, it can cause interference in the donor cells from which the channel is borrowed and prevent future calls in these cells from being completed [21].

As shown in [20], t he SB strategy gives lower blocking probability than static FCA under light and moderate traffic, but static FCA performs better in heavy traffic conditions.

This is due to the fact that in light and moderate traffic condi- tions, borrowing of channels provides a means to serve the fluctuations of offered traffic, and as long as the traffic inten- sity is low the number of donor cells is small. In heavy traffic, the channel borrowing may proliferate to such an extent, due to channel locking, that the channel usage efficiency drops drastically, causing an increase in blocking probability and a decrease in channel utilization [22].

Because the set of borrowable channels in a cell may con- tain more than one candidate channel, the way a channel is selected from the set plays an important role in the perfor- mance of a channel borrowing scheme. The objective of all t h e schemes is t o reduce t h e number of locked channels Call arrival rater of 20-200 callds for each cell

12 IEEE Personal Communications June 1996

(4)

caused by channel borrowing. T h e difference between them is the specific algorithm used for selecting one of the candidate channels for bor- rowing. Along these lines, several variations of the SB strategy have been proposed where chan- nels are borrowed from nonadjacent cells [13, 15, 16, 17, 19, 201. In the following, we discuss briefly each of the proposed schemes.

B o r r o w f r o m t h e Richest (SBR) - I n this scheme, channels that arc candidates for borrow- ing are available channels nominally assigned to one of the adjacent cells of the acceptor cell [lS].

If more than one adjacent cell bas channels avail- able for borrowing, a channel is borrowed from

Simple channel borrowing I

I

Simple borrowing (SB)

I

Borrow from the richest (SBR) Basic algorithm (BA)

Basic algorithm with reassignment (BAR) Borrow first available (BFA)

I

Hybrid channel borrowing

I

. . - ... .

__

the cell with t h e greatest nu mb er of channels

available for borrowing. As discussed earlier, channel borrow- ing can cause channel locking. The SBR scheme does not take channel locking into account when choosing a candidate chan- nel for borrowing.

Basic Algorithm (BA) - This, is an improved version of the SBR strategy which takes channel locking into account when selecting a candidate channel for borrowing [ l S , 161. This scheme tries to minimize the future call blocking probability in the cell that is most affected by the channel borrowing. As in the SBR case, channels that are candidates for borrowing are available channels nominadly assigned to one of the adja- cent cells of the acceptor cell. The algorithm chooses the can- d i d a t e c h a n n e l t h a t maximiizes t h e n u m b e r of available nominal channels in the worst-case nominal cell3 in distance CJ to the acceptor cell.

Basic Algorithm with Reassignment (BAR) - This scheme provides for the transfer of a call from a borrowed channel to a nominal channel whenever a nominal channel becomes available. The choice of the particular borrowed channel to be freed is again made in a manner that minimizes the maximum probability of future call blocking in the cell most affected by the borrowing, as in the BA scheme [16].

Borrow First Available (BFA) - Instead of trying to optimize when borrowing, this algorithm selects t h e first candidate channel it finds [lS]. Here, the philosophy of t h e nominal channel assignment is also different. Instead of assigning channels directly to cells, the channels are divided into sets, and t h e n each set is assigned t o cells at reuse distance 0.

These sets are numbered in sequence. When setting up a call, channel sets are searched in a prescribed sequence to find a candidate channel.

i

I

Simple hybrid borrowing scheme (SHCB) Borrowing with channel ordering ( K O )

Borrowing with directional channel locking (BDCL) Sharing with bias (SHB)

Channel assignment with borrowing and reassignment (CABR)

Ordered dynamic channel assignment with rearrangement (ODCA)

. .. . .. . . . . . . . . . . . .

Performance Comparison - A. general conclusion reached by most studies on the performance comparison of the previous schemes is that adopting a simple test for borrowing (e.g.,

borrowing the first available channel that satisfies the 0 con- straint) yields performance results quite comparable to sys- tems which perform an exhaustive and complex search method to find a candidate channel [13, 15-17]. SBR, BA, and BFA were evaluated by simulation in [15] using a two-dimensional hexagonal cell layout with 360 service channels. The offered load was adjusted for an average blocking of 0.02. The results show that all three schemes exhibit nearly the same average blocking probability versus load with about 25 percent increase in offered load to achieve an average blocking of 0.02. T he BFA has an advantage over the other two in that its comput- ing effort and complexity are significantly less. Here the com- plexity of each algorithm is determined based on the average number of channel tests per call while searching for a candi- date channel t o borrow. In [15], simulation results showed a large variation in the complexity of these algorithms depend- ing on network load. For example, for a 20 percent increase in the traffic, SBR requires SO percent, and the BA 100 percent, more channel tests compared to BFA. A summary of the com- p a r i s o n results bet w een t h e BFA, SBR, BA, a n d BAR schemes is given in Table 2.

Hybrid Channel Borrowing Schemes - In the following we will describe different hybrid channel borrowing schemes.

Simple Hybrid Channel Borrowing Strategy (SHCB) - In the SHCB strategy [5, 13, 171 the set of channels assigned to each cell is divided into two subsets, A (standard) and B (borrow- able) channels. Subset A is nominally assigned in each cell, while subset B is allowed to be lent to neighboring cells. The ratio ] A

I

:

I

B

I

is determined a priori, depending on an estima- tion of the traffic conditions, and can be adapted dynamically in a scheduled or predictive manner [17].

Borrowing with Channel Ordering (BCO) - The BCO, intro- duced in [20] and analyzed in [13, 171, outperforms SHCB by dynamically varying the local t o borrowable channel ratio according to changing traffic conditions [17, 201. In the BCO strategy, all nominal channels are ordered such that the first channel has the highest priority for being assigned to the next local call, and the last channel is given the high- est priority for being borrowed by the neighboring cells. A variation of the BCO strategy, called BCO with reassignment, allows intercel-

- ~. 3 Those cells to which a given channel is nominally assigned are its nominal cells.

Table 2. Companson between BFA, SBR, BA, and BAR.

IEEE Personal Communications June 1996 13

(5)

lular handoff, that is, immediate reallo- cation of a released high-rank channel to a call existing in a lower-rank channel in order to minimize the channel locking effect.

B o rro w i n g with Directional Channel Locking (BDCL) ~ In the BCO strategy, a channel is suitable for borrowing only if it is s i m u l t a n e o u s l y f r e e in t h r e e nearby co-channel cells. This require-

ment is too stringent and decreases the Figure 4. Sharin,q with bias.

Donor cell

A n example is shown in Fig. 4. A call initiated in sector X of cell number 3 can only borrow a channel from set A of the cells numbered 1 and 2.

Channel Assignment with Borrowing a n d R e a s s i g n m e n t (CARB) - T h e CARB scheme proposed in [16] is sta- tistically optimum in a certain min-max sense. Her e channels are borrowed on the basis of causing t h e least harm t o neighboring cells in terms of future call number of channels available for bor-

rowing. In the BDCL strategy, the chan-

nel locking in t h e co-channel cells is restricted t o those directions affected by the borrowing. Thus, the number of channels available for borrowing is greater than that in the BCO strategy. To determine in which case a “locked” channel can be borrowed, “lock directions” a r e specified for each locked channel. The scheme also incorporates reallocation of calls from borrowed to nominal channels and between bor- rowed channels in order to minimize the channel borrowing of future calls, especially t h e multiple-channel borrowing observed during heavy traffic.

Performance Comparison

- As shown by simulation in [13],4 BDCL gives the lowest blocking probability, followed by BCO and F C A , for b o t h u n iform a n d no n unifo rm traffic. T h e reduction of the blocking probability for BDCL and BCO over FCA for the system in [13] is almost uniformly 0.04 and 0.03, respectively, for the range of traffic load tested.

Note that the nonuniform pattern allocation FCA scheme, discussed in the previous section, can be also applied in the case of the hybrid channel borrowing strategies. With the use of nonuniform pattern allocation the relative performance of the BDCL, BCO, and uniform FCA schemes remain the same as before, but the traffic-carrying capacity of a system can be increased by about 10 percent. This advantage is in addition to those gained from the channel borrowing strategies [13]. A summary of the comparison results between the BCO, BDCL, and FCA schemes is given in Table 3.

S h a r i n g with Bias ( S H E ) - In 1231 SHB was proposed: a scheme of channel borrowing with coordinated sectoring. The

-

blocking probability. Likewise, reassign-

ment of borrowed channels is done in a way to cause maximum relief to neighboring cells.

Ordered Channel Assignment Scheme with Rearrangement (ODCA) - Th e OD C A scheme, proposed in [25], combines the merits of CARB and BCO with improvements t o yield higher performance. In ODCA, when a call requests service, the base station of the cell checks to see if there are any nom- inal channels available. If there are channels available, the user will be assigned one on an ordered basis, as in BCO. Here all channels are numbered in predetermined order according to the same criterion as in the CARB scheme, and the lowest- numbered available idle channel is always selected. If all nom- inal channels are busy, t h e cell may borrow a nonstandard channel from a neighboring cell. Once a nonstandard channel is assigned, the availability lists of all affected cells where the assigned channel can cause interference are updated. When- ever a channel is no longer required, the availability lists of the affected cells are updated accordingly. Whenever a stan- dard channel is available, the channel reassignment procedure is initiated to ensure efficient utilization. If there is a nonstan- dard channel in use in the cell, the call served by that channel is switched to the newly freed standard channel; the necessary availability lists are also updated. If no nonstandard channels are used in the cell, a call served by a standard channel with lower priority than the newly freed o n e is switched t o t h e newly freed channel [25].

Performance Comparison

- The performance of ODCA was studied in [25] for a highway microcellular environment with nonuniform tele-traffic load. Performance comparison with SHB strategy is similar to tLe join

biased queue rule [24], which is a simple but effective way to balance the load of servers in the presence of unbalanced traffic. Each cell in the system is divided in three sec- tors,

X,

Y , Z , as shown in Fig. 4.

Only calls initiated in one of these sectors can borrow channels from the two adjacent cells neighboring it ( d o n o r cells). In addition, t h e nominal channels in donor cells are divided in two subsets, A and B , as in the SHCB case. Channels from set A can only be used inside t h e donor cell, while channels in set B can be loaned t o an acceptor cell.

The system in [13] consists of 49 hexag- onal cells, where each cell is allocated 10 channels, and traffic load valyingfrom 20-200 callslh.

T r a f f i c c a r r i e d c a p a c i t y

I

FCA, BCO, SDCL lization compared to the CARB and

I

-

FCA; the ODCA scheme also per-

-...&.

2 -

forms better than CARB and FCA at blocking probabilities below 0.1.

B l o c k i n g p r o b a b i l i t y

i

BCDL, K O , FCA

I +

For example, at a blocking proba-

I bility of 0.05 O D C A is capable of s u m o r t i n e 4 Dercent more traffic Table 3. Comparison betwen BCO, BDCL and FCA.

Channel u t i l i z a t i o n FCA, CARB, ODCA

C o m p u t a t i o n a l c o m p l e x i t y ODCA, CARB, FCA

-+

____ +

T r a E ?Zffied-Ca p a c r t y FCA, CARB, O D X - ,

+ i

!

. -_

Table 4. Compansorz between FCA, CARB, and ODCA

putational overhead in assigning and reassigning channels, and more fre- quent switching of channels due to the reassignment propagation effect.

The performance comparison results between ODCA, CARB, and FCA schemes are summarized in Table 4. Finally, a summary of the com- parison between F C A schemes is given in Table 5.

14 IEEE Personal Communications June 1996

(6)

I... .. . . . . . . . . . . . . . .. .. . . . . . . . ... . . . . . . . . . . . . . . . .. .. . . . .

__

.... .. .. . .

I

i

Hybrid channel borrowing

Dynamic Channel Allocation

Moderate Moderate Better than FCA in light and

I

moderate traffic borrowing in heavy loads

Better than simple channel I

D

ue t o short-term t e m p o r a l and spatial variations of traf- fic in cellular systems, IFCA schemes are not able to attain high channel efficiency. To overcome this, DCA schemes have been stud- ied during the past 20 years. In con- t r a s t t o F C A , t h e r e is n o fixed relationshiu between channels and

Simple FCA

I

Low

lLow I

Better than dynamic and hybrid borrowing in heavy traffic ,

Static borrowina

1

Low-moderate

I

Moderate

I

Better than FCA I

Simple channel borrowing Moderate-high High Better than FCA and static borrowing

I I

in light and moderate traffic

cells in DdA. A11 channels are kept in a central pool and are assigned

dynamically to radio cells as new calls arrive in the system [18, 261. After a call is completed, its channel is returned to the central pool.

In DCA, a channel is eligible for use in any cell provided that signal interference constraints are satisfied. Because, in general, more than one channel might be available in the cen- tral pool to be assigned to a cell that requires a channel, some strategy must be applied to select the assigned channel [lo].

The main idea of all DCA schemes is to evaluate the cost of using each candidate channel, and select t h e o ne with the minimum cost provided that certain interference constraints are satisfied. The selection of 1 he cost function is what differ- entiates DCA schemes [lo].

T h e selected cost function might dep end o n the future blocking probability in the vicinity of the cell, the usage fre- quency of the candidate channel, the reuse distance, channel occupancy distribution under current traffic conditions, radio channel measurements of individual mobile users, or the aver- age blocking probability of the system [22].

Although many claims have been made about the relative performance of each DCA schleme to one or more alternative schemes, the trade-off and thie range of achievable capacity gains are still unclear, and questions remain unanswered: How does each dynamic scheme piroduce its gain? What are the basic trade-offs? Why do some schemes work only under cer- tain traffic patterns? Can different schemes be combined?

What is the value of additional status information of the near-

~. . . ... . . . . . . ._

!

Centralized DCA

I-.

Distributed DCA . .

__

i

_.

8

-j CIR measurement DCA schemes

'

One Dimension Systems

I

Table 6. &namic channel all

by cells? What is the best possible use of the bandwidth [HI?

Based on information used for channel assignment, DCA strategies could be classified either as call-by-call DCA or adap- tive DCA schemes [27]. In the call-by-call DCA, the channel assignment is based only on current channel usage conditions in t h e service a r e a , while in adaptive D C A t h e c h a n n e l assignment is adaptively carried out using information on the previous as well as present channel usage conditions [27, 283.

Finally, DCA schemes can be also divided into centralized and distributed schemes with respect to the type of control they employ. Table 6 gives a list of the proposed DCA schemes.

Centralized DCA Schemes

In centralized DCA schemes, a channel from the central pool is assigned to a call for temporary use by a centralized con- troller. T he difference between these schemes is the specific cost function used for selecting one of the candidate channels for assignment.

First Available (FA) - T he simplest of the DC A schemes is the F A strategy. In F A the first available channel within the reuse distance encountered during a channel search is assigned t o the call. The F A strategy minimizes the system computa- tional time; and, as shown by simulation in [lo] for a linear cellular mobile system, it provides an increase of 20 percent in the total handled traffic compared to FCA for low and mod- erate traffic loads.

. . .- . . . -. -. . . . . . -. . . . . . . . . . . . .. .

First available (FA)

Locally optimized dynamic assignment (LODA) Selection with maximum usage on the reuse ring (RING) Mean square (MSQ)

Nearest neighbor (NN) Nearest neighbor

+

1 (NN

+

1) 1 - clique

i

Locally packing distributed DCA (LP-DDCA) LP-DDCA with ACI constraint

Moving direction (MD)

Sequential channel search (SCS) MSlR

Dynamic channel selection (DCS) Channel segregation

MINMAX

Random minimum interference (RMI)

Random minimum interference with reassignment (RMIR)

Minimum interference (MI)

i

Sequential minimum interference SMI

I

_ - ._

_ .-

.ation schemes.

Locally Optimized Dynamic Assign- m e n t (LODA) - I n t h e L O D A strategy [13, 171 the selected cost function is based on the future block- ing probability in the vicinity of the cell in which a call is initiated.

Channel Reuse Optimization Schemes - T h e objective of any mobile system is t o maximize t h e efficiency of the system. Maximum efficiency is equivalent to maximum utilization of every channel in the system. It is obvious that the short- er the channel reuse distance, the greater the channel reuse over the whole service area. T he cost func- tions s e l e c t e d in t h e following schemes attempt to maximize the efficiency of the system by optimiz- ing t h e reuse of a channel in t h e system area.

Selection with M a x i m u m Usage on the Reuse Ring (RING) - In the

IEEE Personal Communications June 1Y96 15

(7)

R I N G strategy [ l o ] , a candidate channel is selected which is in use in the most cells in the co-channel set. If more than one channel has this maximum usage, an arbitrary selection among such channels is made t o serve the call. If n o n e is available, t h e selection is m a d e based on the FA scheme.

1

Blocking probability Forrcd rermination

-___

____

rate

Channel changing

Carried traffic

I -

NN, MSQ, FA, NN+I

-+

NN

-

I , NN, MSQ, FA

-9

--

NN

+

1, NN, MSQ, FA

.j

NN, NN

+

1, RING, MSQ, FA -3

.-

M e a n Square (MSQ), N e a r e s t UTable 7. Channel reuse optimization schemes Neighbor N N ) Nearest Neighbor

plus O n e ( N N

+

1) - T h e MSQ scheme selects the available channel

that minimizes the mean square of the distance among the cells using the same channel. The NN strategy selects the available channel occupied in the nearest cell in distance t 0, while the

NN +

1 scheme selects an eligible channel occupied in the near- est cell within distance 2 o

+

1 or distance o if an available channel is not found in distance o

+

1 [lo].

Performance Comparison - Computer simulations of FCA, MSQ, NN, and

NN +

1 strategies show that under light traffic conditions,

NN

exhibits the lowest blocking rate, followed by MSQ, FA, and NN

+

1 [27]. Also, the NN

+

1 strategy, when applied to a microcellular system, leads to lower forced call termination and channel changing because the mobile unit is more likely to keep the same channel when i t moves to an adjacent cell [29].

In addition, simulation results of FA, RING, and NN [ l o , 301 show that for both one- and two-dimensional mobile sys- tems, all of the above schemes operate at very low blocking rates until the offered traffic reaches some critical value. A small increase in the offered traffic above this value produces a considerable increase in the blocking probability of new calls and results in very little increase in the traffic carried by the system; the load at which blocking begins to occur in one- dimensional systems [30] is somewhat greater than that in two-dimensional systems [lo]. Finally, the simulation results in [30] show that strategies like R I N G and NN, which use a channel reuse optimization approach, are able to carry 5 per- cent more traffic at a given blocking rate of 3 percent com- pared to a channel assignment strategy like FA, which does not employ any channel reuse optimization. A summary of the performance comparison of the channel reuse optimization schemes is given in Table 7.

1 -Clique - All four previous schemes employ local channel reuse optimization schemes. A global channel reuse optimiza- tion approach is used in the 1-clique strategy. The 1-clique scheme uses a set of graphs, one for each channel, expressing the non-co-channel interference structure over the whole ser- vice area for that channel. In each graph a vertex represents a cell, and cells without co-channel interference are connected with edges. Thus, each graph reflects the results of a possible channel assignment. A channel is assigned from the several possibilities such that as many vertices as possible still remain available after the assignment. This scheme shows a low prob- ability of blocking, b u t when t h e r e a r e a l o t of cells t h e required computational time makes quick channel selection difficult [26].

Schemes with Channel Rearrangement - Compared to FCA schemes, D C A schemes do not carry as much traffic at high blocking rates because they are not able to maximize channel reuse as they serve the randomly offered call attempts. In order to improve the performance of D C A schemes in large

traffic conditions, channel reassign- ment techniques have been suggest- e d [8, 10, 311. T h e basic goal of channel reassignment is t o switch calls already in process, whenever possible, from the channels these calls a r e using t o other channels, with t h e objective of keeping t h e distance bet w een cells using t h e same channel simultaneously to a minimum. Thus, channel reuse is more concentrated, and more traf- fic can be carried per channel at a given blocking rate.

Distributed DCA Schemes

Microcellular systems have shown great potential for capacity improvement in high-density personal communication net- works [2, 32, 33, 341. However, propagation characteristics will be less predictable and network control requirements more intense than in the present systems. Several simulation and analysis results have shown that centralized DCA schemes can produce near-optimum channel allocation, but at the expense of a high centralization overhead [28, 35-38]. Distributed schemes are therefore more attractive for implementation in microcellular systems, due to the simplicity of the assignment algorithm in each base station.

The proposed distributed DCA schemes use either local infor- mation about the current available channels in the cell’s vicini- ty (cell-based) [3942] or signal strength measurements [4345].

In cell-based schemes a channel is allocated to a call by the base station at which the call is initiated. The difference with the centralized approach is that each base station keeps infor- mation about the current available channels in its vicinity. The channel pattern information is updated by exchanging status information between base stations. T he cell-based scheme provides near-optimum channel allocation at the expense of excessive exchange of status information between base sta- tions, especially under heavy traffic loads.

Particularly appealing are the D C A interference adapta- tion schemes that rely on signal strength measurements [43].

In these schemes a base station uses only local information, without the need to communicate with any other base station in the network. Thus, the system is self-organizing, and chan- nels can be placed o r a d d e d everywhere, as n e e d e d , t o increase capacity or to improve radio coverage in a distributed fashion. These schemes allow fast real-time processing and maximal channel packing5 at the expense of increased co- channel interference probability with respect to ongoing calls in adjacent cells, which may lead t o undesirable effects such as interruption, deadlock, and instability.

Cell-Based Distributed DCA Schemes

Local Packing Dynamic Distributed Channel Assignment (LP-DDCA) - I n t h e LP-DDCA scheme proposed in [39], each base station assigns channels to calls using the augment- ed channel occupancy (ACO) matrix, which contains neces- sary and sufficient local information for the base station t o make a channel assignment decision. Let M be the total num- ber of available channels in the system and ki the number of neighboring cells to cell i within the co-channel interference distance. The ACO matrix, as shown in Table 8, has M

+

1

Channel packing refers to the area where a channel cannot be reused and how closely these areas are packed.

16 IEEE Personal Cominunications June 1996

(8)

columns and k,

+

1 rows. The first M columns correspond to the M channcls. The first row indic,ates th e channel occupancy in ~1-11 i and the remaining k , rows indi- cate the channel occupancy pat- tern in the neighborhood of il, as obtained from neighboring base station?. The last column of the matrix corresponds to the number

of current available channels for Table 8. ACO matrix at base station i.

each of t h e ki

+

1 co-channel

cells. Thus, an empty column indicates an idle channel which can be assigned to cell i. When a call requests service from cell i, its basc station uses the ACO matrix and assigns the first channel with an empty column. The content of the ACO table is updated by collecting channel occupancy information from interfering cells. Whenever a change of channel occu- pancy happens in one cell, the base station of the cell informs the base stations of all t h e interfering cells regarding the change in order to update the information in the local ACO matrices.

Adjacent Channel lnterfereiice Constraint - In addition to constraining co-channel interference, the design of a wireless cellular system must also include measures to limit adjacent channel interferencc (ACI). Channel impairments such as crosstalk, premature handoffs, and dropped calls may result from ACI, leading t o d eg radatio n of quality of service.

Although channel filters in both the base station and the mobile unit receivers significantly attenuate signal from adja- cent channels, severe interference may occur in circumstances where the received signal level of an adjacent channel greatly exceeds that of the desired channel. This situation arises often in mobile cellular environments due to the distance differ- ences between th e mobile iunits and th e base stations. To reduce ACI, typical cellular systems employing FCA avoid the use of adjacent channels in the same base station.

All the DCA schemes discussed so far assign channels to calls based on the constraint imposed only by co-channel interference, overlooking ACI. Any of the previous described DCA schemes could be modified so that they assign channels to calls respecting both the minimum co-channel interference and ACI constraints at the expense of a reduction in the total carried traffic.

LP-DDCA with ACI Constraiint - In [lo], a modified version of the LP-DDCA scheme was proposed that incorporates the ACI constraint.

The variation of LP-DDCA imposes additional conditions on th e channel selection from the ACO matrix [40]. If the required channel separation between channels to avoid ACI interference is N,+ the Nadj - 1 columns to the left and right of that channel should have empty entries in the first row of the ACO matrix. When a cad requests service from cell i, its base station searches in the first row of the ACO matrix for a group of 2N,d; - 1 consecutive empty entries where the center column of the group is empty. If successful, it assigns the channel; otherwise, the base station searches for 2N,dj - 1 consecutive empty entries in the first row, where the center columns has only one mark. If a channel is found, it checks to see whether the cell that uses the channel has additional chan- nels available. In that case, it sends a message to the corre- sponding cell, and the bast: station of that cell switches the call using the channel in relation to a new one. Thus, the base station of cell i can usc t h e channel. Otherwise t h e call is blocked.

The simulation results of mod- ified LP-DDCA [40] show t hat when the co-cell channel separa- tion is less than four, which is the case in most r eal systems, t h e impact of t h e additional con- straint on the complexity of the channel selection procedure is insignificant. Also, the fact that modified LP-DDCA is robust to ACT interference is Drimarilv due t o its ability t o pr ot i d e fl e kbl e reuse packing of channels by allowing up t o one local reas- signment to accommodate a new call.

Moving Direction (MD) - The MD strategy was proposed in [41, 421 for one-dimensional microcellular systems. In these systems, forced call termination and channel changing occur frequently because of their small cell size [42]. The MD strat- egy uses information on moving directions of the mobile units to decrease both the forced call termination blocking proba- bility and t h e channel changing. An available channel is selected among those assigned to mobile units that are else- where in the service area and moving in the same direction as the mobile in question. The search for such a channel starts from the nearest noninterfering cell to the one where the new call was initiated, and stops at the cell that is a reuse dis- tances away, where a is a parameter.

A channel assignment example is given in Fig. 5 where b, c, d, and e are the available channels, and DR is the minimum reuse distance. For this example the parameter a is set to one.

T h e new call a t t e m p t is assigned channel b because t h e mobile requesting the channel is moving in the same direction as the mobile in cell number 5.

T h e sets of mobiles moving in t he same direction and assigned the same channel are thus formed. Thus, when a mobile of a set crosses a cell boundary, it is likely that a same- set of mobiles has already crossed out of its cell to the next cell. In this manner, a mobile can use the same channel after handoff with higher probability. This lowers the probability of both changing channels and forced call termination. The strat- egy is efficient in systems where mobiles move at nearly the same speed through the cells laid along a road or a highway and for one-dimensional microcellular systems.

The simulation results in [4%] for a one-dimensional system show th at t he M D strategy provides lower probability of forced call termination compared to the NN,

NN +

1, and FCA strategies. Although the MD scheme has attractive fea- tures, it is not obvious how it could be expanded to a two- dimensional system. A summary of the comparison results is given in Table 9.

Signal Strength Measurement-Based Distributed DCA Schemes - A large body of research has been published on the performance analysis of channel allocation schemes, both FCA and DC A [3, 5, 12, 46, 471, in which knowledge of the mobiles’ locations is not taken into account. In all of these schemes, channels are allocated to cells based on the assump- tion that t h e mobile may b e located anywhere within t he boundary of the cell. Thus, t he packing of channels is not maximal. These schemes suffer from the fact that the selected fixed reusability distance might be too pessimistic.

In the interference adaptation schemes, mobiles measure t h e amount of co-channel interference t o determine t h e reusability of the channel. If a mechanism is assumed to exist by which mobiles and base stations can measure the amount of interference, as was done in [48], then maximal channel

IEEE Personal Communications June 1996 17

(9)

. - _ _ - Cali atJcmpt

? 4 5

I

I t - - -

DR - - + 1

(t - .- - D R T l - - - -.+

I

I . - - - - - -

Figure 5 . il4oving direction strategy illustration

packing could be achieved. An example of a system based on this principle is the Digital European Cordless Telecommuni- cations (DECT) standard [49].

However, local decisions can lead to suboptimal allocation.

In interference adaptation DCA schemes, mobiles and base stations estimate CIR and allocate a channel to a call when predicted CIRs are above a threshold. It is possible that this allocation will cause the CIR of established calls to deterio- rate, in which case a sewice interrupt occurs. If the interrupted call cannot find an acceptable new channel immediately, the result is a premature service termination, referred to as dead- lock. Even if the interrupted call finds an acceptable channel, setting up a link using the new channel can cause interruption of another established link. These successive interruptions are referred as instability. If no channel is available for the initial call request, the call is blocked [43, 501

Sequential Channel Search (SCS) - T he simplest scheme among the interference adaptation DCA schemes is the SCS strategy [43], where all mobileibase station pairs examine channels in the same order and choose the first available with acceptable CIR. It is expected that SCS will support a volume of traffic by suboptimal channel packing at the expense of causing many interruptions.

Minimum Signal-to-Noise Interference Ratio (MSIR) - In MSIR [43], a base station searches for the channel with the minimum interference ratio in the uplink direction. Because it first assigns unused or lightly loaded channels to new calls, MSIR has a relatively lower interruption probability than SCS; on the other hand, it is more vulnerable to blocking than SCS. It is generally observed by the simulation results that there is a trade-off between the goals of avoiding call blocking and avoiding interruptions [43].

Dynamic Channel Selection (DCS) - DCS: as presented in [Sl J, is a fully distributed algorithm for flexible mobile cellular radio resource sharing based on the assumption that mobiles are able to measure the amount of interference they experi- ence in each channel. In DCS, each mobile station estimates the interference probability and selects the base station which minimizes its value. The interference probability is a function of a number of parameters, such as the received signal power from base stations, the availability of channels, and co-channel interference. In order to evaluate the interference probability, specific models for each of the above parameters should be developed. In 1701, models are developed to calculate proba- bilities of channel availability, desired carrier power, and the CIR for constant traffic load.

Channel Segregation - The channel segregation strategy was proposed in [44, 451 as a self-organized dynamic channel assignment scheme. By scanning all channels, each cell selects a vacant channel with an acceptable co-channel interference

level. T he scanning order is formed independently for each cell in accordance with the probability of channel selectabili- ty, P(i), which is renewed by learning 1441. For every channel i in t h e system, each cell keeps the current value of P ( i ) . When a call request arrives at the base station, the base sta- tion channel with the highest value of P(i) under observation is selected. Subsequently, the received power level of the selected channel is measured in order to determine whether the channel is used or not. If the measured power level is below (or above) a threshold value, the channel is deter- mined to be idle (or busy). If the channel is idle, the base sta- tion starts communication using the channel, and its priority is increased. If the channel is busy, the priority of the channel is decreased and the next-highest-priority channel tried. If all channels are busy, the call is blocked [44, 451. The value of P(i) and the update mechanism determine the performance of the algorithm. In [44], P ( i ) is updated to show the successful transmission probability on channel i as follows:

P(i) = [P(i)N(i)

+

l ] / [ N ( i )

+

11 and N ( i ) = N ( i )

+

1 if the channel is idle P(i) =

[P(i)N(i)]/[N(i) +

1 j and N ( i ) = N ( i ) -t 1 if the channel is busy

Here N ( i ) is the number of times channel i is accessed. In [45] t h e updat e mechanism f or P ( i ) is defined as P ( i ) =

N3(i)/N(i),

where

N,(i)

is t he number of successful uses of channel i.

Because no channel is fixed to any specific cell, channel segregation is a dynamic channel assignment method. It is also autonomous, for no channel reuse planning is required and it is adaptive to changes in the mobile environment 1451.

The simulation results in [44] show that the channel segrega- tion scheme uses channels efficiently and decreases the num- ber of intracell handoffs, that is, the reassignment of channels to avoid interference. It also decreases the load of the switch- ing system as well as quality degradation during a handoff period [44]. Simulation results show that interference due to carrier sense error is reduced by 1/10-1/100 with channel seg- regation [44]. Also, the blocking probability is greatly reduced compared to FCA and DCA schemes. Speed of convergence to the optimum global channel allocation is an important issue in implementing channel segregation. Based on the anal- ysis in 1441, channel segregation quickly reaches some subopti- mal allocation, b ut convergence t o t h e opt i m um global allocation takes a prohibitively large amount of time because there are many local optimum allocations.

The discussion in 1451 shows that channel segregation can be successfully applied to a TD multiple access/FD multiple access (TDMAIFDMA) or multicarrier TDMA system. As discussed in [45], the difference in the performance of the FDMA and TDMA/ FDMA systems using channel segrega- tion is small, and one-carrier T DMA and FD MA have, in principle, similar performance. The advantages of channel seg- regation are summarized in Table 10.

( 3 )

One- Dimensional Cellular Systems

All the DDCA schemes described in this section are applica- ble for one-dimensional cellular mobile systems. One-dimen- sional structures can be identified in cases such as streets with tall buildings shielding interference on either side [SO].

Minimum lnferference (Ml) - The MI scheme is well known and among the simplest for one-dimensional cellular systems.

It is incorporated in the Enhanced Cordless Telephone (CT- 2) and DECT systems 1501. We present here the MI and its modifications.

In an MI scheme, a mobile signals its need for a channel to

18 IEEE Personal Communications June 1996

參考文獻

相關文件

6 《中論·觀因緣品》,《佛藏要籍選刊》第 9 冊,上海古籍出版社 1994 年版,第 1

‰ For example Competition Provisions in Hong Kong Chapter 106 Telecommunication Ordinances 7K : S.1 A licensee (entity) shall not engage in conduct which, in the opinion of

In case of non UPnP AV scenario, any application (acting as a Control Point) can invoke the QosManager service for setting up the Quality of Service for a particular traffic..

Please liaise with the officer in your school who are responsible for the Class and Subject Details Survey for using of the same class names in both the Class and Subject

• When a call is exercised, the holder pays the strike price in exchange for the stock.. • When a put is exercised, the holder receives from the writer the strike price in exchange

• When a call is exercised, the holder pays the strike price in exchange for the stock.. • When a put is exercised, the holder receives from the writer the strike price in exchange

Provide all public sector schools with Wi-Fi coverage to enhance learning through the use of mobile computing devices, in preparation for the launch of the fourth IT in

Provide all public sector schools with Wi-Fi coverage to enhance learning through the use of mobile computing devices, in preparation for the launch of the fourth IT in