Вы находитесь на странице: 1из 8

Ad hoe

Rusty O. Baldwin Nathaniel J. Davis IV Scott E Midkiff


baMwinr@ vt. edu ndavis @ vt. edu midkiff@ vt. edu
Center for Wireless Telecommunications, Bradley Department of Electrical and Computer
Engineering, Virginia Polytechnic Institute and State University,
Blacksburg, VA

We develop and analyze a simple, elegant medium access control (MAC) protocol for use inn trans-
mitting real-time data in point to point ad hoc wireless local area networks (WLANs). Our en-
hancement of IEEE 802.11, real-time MAC (RT-MAC), achieves dramatic reductions in mean delay,
missed deadlines, and packet collisions by selectively discarding packets' and sharing station state
information. For example, in a 50 station network with a normalized offered load of O.7, mean delay
is reduced from more than 14 seconds to less than 45 ms, late packets are reduced from 76% to
less than 1%, and packet collisions are reduced from 36% to less than 1%. Regression models are
developed from simulation data to describe network behavior in terms of throughput, mean delay,
ratio of late packets, and ratio of collisions. Stations using RT-MAC are interoperable with stations
using IEEE 802.11.

I. Introduction typically expensive and not easily transferred to other ap-


plication areas. Given the increasing demand for real-time
Communications pervades modern society. From broadcast systems, especially in the areas of voice and video data, a
radio and television, to the exchange of information via two- low-cost solution to real-time communications is highly de-
way radios, telephones, cellular phones and pagers, to the sirable. IEEE 802.11 is a recent (1997) standard developed for
global internet, communications pervades every aspect of our WLANs [14]. It has capabilities that can be exploited to pro-
lives. It has also had an enormous impact on the industrial vide real-time service. A standards-based solution offers the
and manufacturing industries. Production lines and industrial potential for a low-cost implementation of an effective real-
control systems rely more and more on computers, often sev- time system. Additionally, 1EEE 802.11 supports probabilistic
eral, to control manufacturing processes and robotic assembly access to the medium which can be used to support a wide
systems. These computer systems in turn require timely (i.e., range of application areas.
real-time) information via communication networks to coor- The remainder of this paper is organized as follows. Sec-
dinate their actions. Timely information is vital to the suc- tion II briefly describes IEEE 802.11. Section III describes
cess of military operations as well. Modern warfare requires the modified IEEE 802.11 protocol, RT-MAC. Section IV ex-
extensive communication between the military services (e.g., plains the simulation model, methodology, and assumptions
the Army, Air Force, Navy, etc.) as well as between units in used. Section V presents an evaluation of RT-MAC and re-
those services. Moreover, military personnel in the field need gression models derived from simulation data. Conclusions
to communicate with their weapon systems which are often and future work are discussed in Section VI.
controlled remotely. To provide this timely information, new
protocols must be developed which support the transmission of
real-time data such as voice, video, and automatic control in- II. IEEE 802.11
formation within a given amount of time. This paper describes
a novel protocol used to support such data transmission in the The IEEE 802.11 standard defines both a multiple access con-
context of a point to point ad hoc wireless local area network trol (MAC) protocol and physical layer implementations. At
(WLAN). the MAC layer, IEEE 802.11 supports both infrastructure and
From the earliest wireless LANs such as ALOHA [1], re- ad hoc networks. In an infrastructure network, stations are
search into wireless LANs has continued uninterrupted. Early granted access to the medium by a station known as the point
research identified fundamental principles and analysis tech- coordinator (PC) and use the Point Coordination Function
niques [2], [16], [18], [19], [201, [21], [221, [33] which are (PCF) for packet transmission. Due to the overhead in this cen-
still applicable, as well as fundamental problems that are still tralized access scheme, PCF have been regarded as unsuitable
encountered [31], [32]. Much, if not most, of past research for real-time data [29], [34]. In an ad hoc network, stations
has been focused on increasing throughput and reducing mean "compete, for access to the medium using the Distributed Co-
delay. More recently, a measure of attention has turned to ordination Function (DCF). Both infrastructure and the ad hoc
the area of real-time WLANs where individual packet delivery networks ultimately use DCF for medium access. The DCF
times are the foremost concern [7], [10], [24], [26], [29], [36]. prioritizes access to the medium by specifying a time interval
Excellent surveys of work in real-time LANs can be found in between frames known as the inter-frame space (IFS). By def-
[23], [271. inition, during an IFS the medium is idle. The different types
Most real-time systems are specialized; designed and built of IFSs, along with the backoff mechanism described below,
to satisfy a unique requirement. As such, these systems are allows a station to determine whether it may transmit. It is

20 Mobile Computing and Communications Review, Volume 3, Number 2


its deadline; collision avoidance is achieved by deferring BV
Immediateaccess when decrements while the medium is busy and by doubling CW
medium is free > DIFS
upon transmission failure as described in Section II.
RT-MAC uses two additional pieces of information to
achieve its result: a transmission deadline and the transmit-
ting station's next BV. When a real-time packet is submitted
~ N e x t Packet for transmission, a transmission deadline (i.e., when transmis-
i Slot Time sion of the packet must begin by) is associated with the packet.
iSelectslot and decrementbackoff This value is only needed until the packet is either successfully
as long as mediumis idle
transmitted or discarded and therefore does not become part of
the packet itself. When the packet is removed from the queue
Figure 1: IEEE 802.11 Basic Access Method
for transmission the next BV to be used by the transmitting
station (i.e., the number of idle slots the station will wait prior
also known as the basic access method. to transmitting again) is placed in the packet header. Stations
that hear the transmission will use this BV to avoid selecting
There are four types of IFSs: Short IFS (SIFS), PCF IFS
the same BV for their backoff timer. This will be described in
(PIFS), DCF IFS (DIFS), and Extended IFS (EIFS). EIFS,
detail in Section III.B. Note that BV is NOT the value of CW
which is the longest IFS in terms of time, is used when bit
(cf., Section II), but rather the result of the random selection
errors introduced by the physical medium cannot be corrected
of a BV from the range of [0, CW-1].
by the radio receiver. Transmission after SIFS, the shortest
IFS, is reserved for the PC to send any type of frame required
or for other stations to begin transmission of an acknowledg- IILA. Transmission Control
ment (ACK) frame, a clear-to-send (CTS) frame, to respond
to polling by the PC, or to send a fragmented MAC protocol The transmission deadline of a packet is examined at three key
data unit (MPDU). Similarly, access after PIFS is reserved for points to determine whether to discard the packet. By dis-
stations to begin transmission of PCF traffic. After DIFS, if carding a packet as soon as possible after determining that its
a station determines that the medium is idle, it may transmit a deadline has been exceeded, transmission queue throughput is
pending packet. If the medium is not idle after DIFS, a backoff increased and as a result, the likelihood that other packets in
timer is set by selecting a random integer (hereafter known as the queue will meet their deadlines is increased. The examina-
the backoff value (BV)) from a uniform distribution over the tion points (described below) were chosen because they follow
interval [0, CW-1 ], where CW is the width (in slots) of the con- unpredictable delays that a packet suffers prior to transmission.
tention window. This BV is the number of idle slots the station A packet is first examined when it is removed from the
must wait until it is allowed to transmit. For every idle slot de- transmission queue in preparation for transmission. If the
tected (after a DIFS), the timer is decremented by one. If the packet has already exceeded its transmission deadline, it is
medium becomes busy prior to the timer expiring, the timer is discarded and the next eligible packet in the queue (if any)
frozen until the next DIFS, upon which the timer decrements is selected. At this point, the station may need to wait for the
again. Upon expiring, the station transmits its packet. If there backoff timer to expire. During this time, other stations could
is a collision, CW is doubled until it reaches a predefined max- possibly transmit. After the backofftimer expires, the packet is
imum value, CWmax. Upon a successful transmission, CW is examined again. If the packet deadline has been exceeded the
reset to the default minimum value of CWmin. Figure 1 [14] packet is discarded, otherwise, it is transmitted. Assuming the
shows the structure of the basic access method. transmission is successful, the next eligible packet is selected
and the process repeats. If the transmission is not successful
At the physical layer, IEEE 802.11 uses carrier sensing to
(that is, no acknowledgement is received), the packet deadline
detect an active channel. It supports a 1 Mbps and a 2 Mbps
is again examined and the packet is discarded if the deadline
data rate. Recently, proposals to support data rates up to
has been exceeded. If the deadline has not yet been exceeded,
30 Mbps as well as data encoding to enhance the coverage
the packet is submitted for retransmission. Using this trans-
area has been introduced [9]. The interested reader is encour-
mission control (TC) algorithm, a packet that is successfully
aged to consult [8], [12], [14], [34], and [35] for more detailed
received will never be late. The TC algorithm is summarized
information on IEEE 802.11.
in Figure 2.

III. RT-MAC Description III.B. Enhanced Collision Avoidance

Two major factors that impact the ability of a real-time WLAN The enhanced collision avoidance (ECA) algorithm has two
to meet packet deadlines are the transmission of packets that components. First, rather than employ a static value for CW,
have already missed their deadlines and packet collisions. CW is set to be eight times the number of stations in the net-
Packets that have missed their deadlines are assumed to be work (i.e., CW = 8N). If a static CW value is used, the proba-
unusable by the receiving station so transmitting them consti- bility of two or more stations choosing the same BV increases
tutes a double failure. The first failure is the missed deadline with N. By making CW a function of N, this probability does
itself, the other is the wasted channel capacity that could have not increase. N is assumed to be estimated either by track-
been used to transmit a usable packet. IEEE 802.11 does not ing the number of unique station addresses that have transmit-
have any means of detecting whether a packet has exceeded ted over the last z seconds where z is a suitable value, or by

Mobile Computing and Communications Review, Volume 3, Number 2 ~.. 21


F +
Discard
Packet

t R e m o v e Packet
Prom Q u e u e
Discard
Packet

(yes)

(no)

RT = R e a l - t i m e V---
R B V = Received Backoft Value
(yes)
C B V = Curreat Backoff Value

Figure 3: Enhanced Collision Avoidance (ECA) Algorithm

the receiving stations current BV. If a suitable value cannot be


found, the range of values will be doubled (i.e., [0, 2CBV-1])
until a suitable value can be found. Figure 3 summarizes the
second component of the ECA algorithm.

iV. Simulation Model


To investigate the performance of our protocol, a simulation
model of the DCF (cf., Section II) of IEEE 802.11 was devel-
Figure 2: Transmission Control (TC) Algorithm oped and implemented using the network simulation tool OP-
NET version 3.5A [28]. The model was developed using the
Specification and Description Language (SDL) [ 15] specifica-
a method such as the one described in [8] where N is esti- tion of IEEE 802.11 that is found in Annex C of the standard.
mated as a function of channel load. The value of 8 is loosely The model was validated against the results obtained in [8].
based on the CW equation used in [8]. In [8], this value is Because [8] was based on an earlier draft, several of that pa-
determined based on packet transmission time and by estimat- per's parameters did not match those in the latest IEEE 802.11
ing channel utilization over an arbitrary observation period. standard. For the validation, we changed the values of these
Rather than estimating these parameters, we use a fixed value parameters in our model to match [8]. For a more detailed dis-
of 8 and rely on the second component of our algorithm (de- cussion of the model construction and validation readers are
scribed below) to resolve any BV conflicts that may still occur.
encouraged to consult [4].
Several schemes that dynamically alter the value of CW have The network under investigation is a fully meshed ad hoc
been proposed and can be found in [7], [8], and [10]. network of 5 to 50 homogeneous stations, which is consistent
The Second component of the ECA algorithm involves ad- with other studies [7], [8], [10]. The physical layer is assumed
vertising the transmitting station's next BV and tracking the to be a Direct Sequence Spread Spectrum (DSSS) transmission
BVs of other stations in the network. Prior to transmitting a system. The remainder of this section describes the simulation
packet, a station will select its BV from the range of [0, CW- parameters, factors, and response variables.
1], excluding BVs that are known to be in use. This selected
BV will be the one used following the current transmission. It
IV.A. S i m u l a t i o n P a r a m e t e r s
will also be placed in the packet header and transmitted along
with the packet. Stations that hear the transmission will place Several parameters are used to control the simulations. The
the BV in a table of BVs "in use". During idle slots, a sta- simulation model can vary over 30 different parameters, most
tion will decrement its own BV (as in IEEE 802.11) as well as of which are IEEE 802.11 parameters. The most significant
every BV in its table of BVs. parameters are listed in Table 1. Any other parameters not
A station may receive a packet that indicates the sending listed are either not implemented in the model or are set to the
station has chosen the same BV as the receiving station. This default value for a DSSS system.
could occur due to new stations joining the network or due
to BVs not being received because of collisions or bit errors.
IV.B. S i m u l a t i o n F a c t o r s
In such cases, the receiving station chooses another BV since
a collision will certainly occur (assuming both stations have Factors are parameters that are varied during the simulation
a packet to transmit). To prevent a station that must choose such that they significantly impact system performance when
a new BV from being unduly penalized, the new BV is cho- altered [17]. The set of factors used include the number of
sen (if possible) from the range of [0, CBV-1] where CBV is stations, N, the total offered load, G, the channel model, E,

22 Mobile Computing and Communications Review, Volume 3, Number 2


Tabte 1: Simulation Parameters Table 3: Simulation Factors
Parameter Value [ Factor 'Values
Channel Bit Rate
CWmin
CWmax
Slot Time
SIFS
1 Mbps
31"*
1023"*
20 # s
50 # s
[ Number of Stations ( N )
Total Offered Load ( G )
Channel Model (/5)
Protocol
5, I0,20.30,40,50
0.3 0.5,0.7,0.9
0 (Idesd), I (Bursty)
[EEE 802.11 RT-MAC

DIFS Calculated at runtime


ACK Length Calculated at runtime
PHY Header Calculated at runtime VLC. Response Variables
ACK Timeout Calculated at runtime
RxTx _Turnaround_Time 5 ~s
** IEEE80211 oaly
Four response variables are of interest. Of primary interest,
given the real-time emphasis, is the missed deadline ratio, F.
The missed deadline ratio is number of packets that exceed
their deadline over the number of packets removed from the
Table 2: Traffic Models
Factor Value queue for transmission (cf., Figure 2). A discarded packet (due
tnterarrival Distribution Constant to exceeding the transmission attempt count or due to the TC
Deadline Distribution Constant
(same as interarrival time)
algorithm) is deemed to have exceeded its deadline. A second
Packet Size (bytes) 83 response variable is the collision ratio, C. The collision ratio
Discarded Packets Resubmitted 0%
-X~iT~ics Interarrival Distribution Poisson
is the number of packet collisions over the number of trans-
Deadline Distribution Truncated Normal mission attempts. The third response variable is the aggregate
Mean = 380 ms, Min=21 ms
system throughput, S. Finally, the fourth response variable is
Max= 1 s
Packet Size (bytes) 775 the mean delay of successfully received packets, D. The sub-
Discarded Packets Resubraitted 50% scripts Std and RT, when used with the response variables,
mean the IEEE 802.11 and RT-MAC protocols, respectively.

where E = 0 means an ideal channel and E = 1 means V. Performance Analysis


an errored channel, and the protocol in use (IEEE 802.11 or
RT-MAC). We investigate two types of traffic: telemetry and A full factorial experiment with replications was chosen for
avionics. The telemetry traffic model is representative of traf- this effort. At least five replications were needed to obtain
fic on a MIL-STD-1553B serial data bus [3], while the avionics suitable confidence intervals for the response variables [17],
traffic model is representative of the requirements for traffic on [25]. A simulation run was terminated when the confidence
the avionics bus on the Boeing 777 [11]. interval widths of all four response variables were within a
The telemetry model has short packets and constant interar- given percentage of their mean values [17] or after a simula-
rival times such as can be found in a cyclic real-time system. It tion time of 100s (telemetry traffic model) or 1000s (avionics
is intended to be a "worst-case" traffic model in the sense that traffic model). This termination criteria helped to ensure that
short packets induce greater overhead as well as more oppor- the system was in steady-state. The throughput and mean de-
tunity for collisions. The transmission deadline for the packets lay confidence interval widths were required to be less than or
is equal to the interarrival time. The total offered load, G, is equal to 2% of their mean value and the missed deadline and
varied by adjusting the interarrival time appropriately. Pack- collision ratio confidence interval widths were required to be
ets discarded by the TC algorithm are assumed useless to the less than or equal to 10% of their mean value. Data collection
receiver and are not resubmitted for transmission. began after 5s of simulation time: Using the factors in Tables 2
and 3 and performing five replications results in a total of 960
The avionics traffic model has moderately sized packets
simulation runs. Individual simulations runs were performed
which arrive according to a Poisson process. Deadlines are
on several machines simultaneously. They took the equivalent
normally distributed. The total offered load, G, is varied by
of approximately 8000 CPU hours on a Sun UltraSparc 2 with
adjusting the mean interarrival time appropriately. In this traf-
256 MB of RAM a 170 MHz processor running Solaris 2.5.
fic model, some of the packets discarded by the TC algorithm
are assumed to still be useful if transmitted. Therefore, 50%
of discarded packets are randomly resubmitted to the transmis- V.A. Simulation Results
sion queue. Table 2 provides details of the two traffic models. We found dramatic and noteworthy performance improve-
The other factors that were varied and their values are shown ments using RT-MAC. In general, for both the telemetry and
in Table 3. The channel model used in the simulations was ra- avionics traffic models, RT-MAC stabilized the behavior of the
ther ideal (E = 0) or an errored channel (E = 1) using a two- response variables whereas IEEE 802.11 was characterized by
state bursty error model as in [51, [6], and [13]. The errored asymptotically increasing mean delay, collisions, and missed
channel is in a "good" state for an exponentially distributed deadlines beyond a certain total offered load. Figure 4 is a case
amount of time with mean 5.0s. In the good state, there are no in point.
bit errors. It is in a "bad" state for an exponentially distributed Figure 4 shows Dstd and DnT using the telemetry traffic
amount of time with mean 0.1 s. In the bad state, the probabil- model and an ideal channel. Dstd increases asymptotically as
ity of a bit error is 0.8. The error probability and state times are G increases from 0.3 to 0.5. DnT actually decreases as G in-
based on the observed behavior of real WLANs [5], [131. The creases. This behavior of RT-MAC can be attributed to the TC
average BER varies with G but is typically quite poor, with algorithm. Since late packets are discarded rather than trans-
B E R ~ 2 x 10 . 2 . mitted, the throughput of packets in the transmission queue is

Mobile Computing and Communications Review, Volume 3, Number 2 ........... 23


100 - IEEE 802.11 .........
RT-MAC Table 4: Telemetry Traffic Model Simulation Results
Response Range oF-Response 90% Confidence Interval
"g" 10 Variable Variable Values is l e s s than or equal to
N 550 : ......... ::::::::::::::::::::::::::::
N = 40 - - ~fi~'~-2222222222222252222222222222222222
Throughput 0.28 _< Sstd < 0.33 _+0.0033
5' = 3 0 - - %",', . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
N: 20- xgV/ 0.28 < S R T < 0.34 ±0.0032
N = 10 - - : , z Z 2 v L ¢ , . . . . . . . . . . }--27,- .................... Delay (sec) 0.006 < D s t d < 17.0 ± 0 . 0 0 2 7 , D s t d < 1.70
.%',,, ," ," ///N = 40 ± O . 0 7 1 , D s t d > 1.70
. . ".;-> . ' . - " ///~\, : ao 0.002 _< D R T _< 0 . 0 4 3 i0.00021
,',,;'," .'," N = 2o M i s s e d Dead- 0.01 < / / S t d _< 1.00 ~0.0038
0.1- ..'.?/..'." N lo =

line R a t i o 0.0 < F R T _< 0.63 ±0.0051


Collision 0.03 < C6'td _< 0.40 :i:0.0045
e~ Ratio 0.01 ~ CRT ~ 0.04 ~0.0025
,~ 0.01-

G=0.9
I I.... I t-- I -- I I
IEEE 8 0 2 , 1 1 . . . . . - .......................
0.3 0.4 0,5 0.6 0.7 0.8 0.9
Total Offered Load (G) RT-MAC / G = 0.7
/ G:0.9
•2 0.1
~o
/"
Figure 4: Mean Delay -Telemetry Traffic Model g
--'-'''-- C=0.7
O.Ol /
0.4- IEEE 802.11 ...... /'

0201- / G : 0.5 /

0.3- G = 0.5,0.7,0.9 .~'~" .*'" G ~ 0.3


7,
10 4_
g
- - I 7 - - I I I
0.2- 10 20 30 40 50
,"" ~~ ' ' " G = 0.3 Stations (N)

0.1-
Figure 6: Missed Deadline Ratio - Avionics Traffic Model
G : 0.30.9

For the telemetry traffic model, the range of response vari-


0- I T If ~ t
I0 20 30 40 50 able values is shown in Table 4. Note that the minimum
Stations (N) and maximum values shown in the tables do not necessar-
ily correspond to the minimum load/station or the maximum
Figure 5: Collision Ratio - Telemetry Traffic Model load/station network configurations. Using the regression
models in Section V.B, the behavior of particular network
configurations can be determined. Each of the response vari-
much greater than that of IEEE 802.11 and consequently the
able samples had a slightly different confidence interval width,
mean delay decreases. In the simulation model, the transmis- therefore, the maximum confidence interval width is shown in
sion queue was limited to 200 packets. Using IEEE 802.11, the table.
as G increased from 0.5 to 0.9, the number of packets blocked Comparable results were obtained for the avionics traffic
due to a full transmission queue increased from 10% to 55% model. Using this traffic model, IEEE 802.11 performed fairly
of arriving packets. For RT-MAC, the queue length never ex- well for G < 0.5 and N < 20. For other network configu-
ceeded 10 packets. For the errored channel, the IEEE 802.11 rations, behavior in terms of D, F , and C was similar to the
network typically experienced a 10 to 250 ms increase in mean telemetry model.
delay. The RT-MAC network had no increase in mean delay
Figure 6 shows Fsta and FRy using the avionics traffic
since the packets that became late due to retransmissions were model and an ideal channel. Fsta increased asymptotically as
discarded by the TC algorithm.
G increased from 0.7 to 0.9 for N < 20. FRT was much more
For an ideal channel, Fsta was close to zero for G = 0.3, stable, never exceeding 0.14 even in worst-case conditions.
N < 40 and reached 1.0 for G > 0.5. FRT increased ap- For the avionics traffic model, C and D also showed behav-
proximately linearly from 0.0 to 0.63 (independent of N ) as G ior comparable to the telemetry traffic model. For the avionics
increased. traffic model, the range of response variable values is shown
Packet collisions, C, showed similar behavior. Figure 5 in Table 5. As with the telemetry table, the minimum and
compares the packet collision ratios of IEEE 802.11 and RT- maximum values shown do not necessarily correspond to the
MAC. Csta is characterized by rapidly increasing collision ra- minimum load/station or the maximum load/station network
tios as G increases from 0.3 to 0.5 and a maximum collision configurations and the maximum confidence interval width is
ratio strongly influenced by N. CRT shows extremely stable shown. The increased throughput can primarily be attributed
behavior. It never exceeds 0.045. to the larger packets sizes.
Aggregate throughput, S, for both protocols was generally Other simulation studies were conducted to determine the
about 0.3. Usable throughput, U, (the throughput of packets relative benefit of the different components of RT-MAC. To
that arrive prior to deadline expiration where U = S(1 - F ) ) determine this, we ran simulations (with 5 replications) using
was 0.0 for Ustd for any G ~ 0.5 while URT achieved 0.13 to a network with an ideal channel, avionics traffic model, and
0.23 for similar network configurations. N = 40, G = 0.7. Figure 7 gummafize the results. In terms

24 Mobile Computing and Communications Review, Volume3, Number 2


Table 5: Avionics Traffic Model Simulation Results Table 6: Telemetry Traffic Model Regression Models
Response Range of" Response 90% Confidence Interval
Variable Variable Values is less than or equal to Variable Regression Model
Throughput 0.29 < SStd ~ 0 , 7 6 ~0.0070 Throughput S$td : 5 . 2 9 2 x 10 - a G a N -- 6 . t 5 6 × I O - a G U N 0.92
0,29 _< ,5'RT < 0 . 8 3 ±0.0039
--0.204(72 + 0 . 3 0 5 G + 0.222
Delay(sec) i 0 . 0 0 9 < Ds'td < 36.7 ~ 0 . 0 0 2 8 , [_)st d ~ 4.20
S R T : - - 2 . 2 1 5 x 10 ~ G N ~ + 1.782 × i O - 4 G N ~ 0.91
-~0,028, DStd > 4.20
--3.091 × 1 0 - 3 G N + 1 . 3 9 4 G 3 -- 2 . 8 2 1 G 2
0 . 0 0 9 ~ Dt~T < 0 . 1 3 4 ±0,00090
+ 1 . 7 9 6 G - 0.0406
Missed Dead- 0.0 < F.vt~i _< 0.99 ±0.0038
Delay (sec) Ostd : (-5.005 × i O - ~ G ~ N 0.99
line Ratio 0.0 < F R T < 0.14 ::t=0.0028
+ 2 . 8 1 0 × 210 6 G N a
Collision 0.002 < C s t d _< 0 . 4 0 ±0.0055
Ratio 0 . 0 0 0 8 <_ (TRT <_ 0.03 ~0.0015 --3.117 X 1 0 - 4 G N 2
+ 1 . 6 8 2 x 10 - 2 G N + 4 . 8 7 0 G 3
-- 1 0 , 1 8 7 G z + 6 . 8 6 6 G - 0.499) 2°
s ...................................................... "
DR7, : --8.048 × I O - ~ G N + 1.080 x 10 a N 0.99
0.7
q-3.002 × 10 - 4
Missed /~\~ta = s i n ~ ( 2 7 - 5 7 4 G a -- 5 8 . 0 5 5 G z 0.99
0,6 Deadline + 3 9 . 6 7 0 G + 1,1.80 × 1 0 - 6 N a
Ratio - 2 . 6 2 5 × 1 0 - a N - 7.186)
0.5 FRT = - - 0 . S 0 7 G ~ + 1 . 9 9 3 ( / -- 0.521 0.99
Collision C s t a = - - 0 . 4 3 7 G 2 + 0 . 6 2 8 G - :l.015 × 1 0 - ' ~ N z 0.98
0.4 Ratio + 0 . 0 1 2 6 N - 0.184
C R T = 5.445 × 1 0 - ' N a -- 6.258 x 1 0 - ~ N ~ 0.91
+ 2 . 3 7 8 × 1 0 - a N + 9,898 × 10 - a
0.3

0.2

Table 7: Avionics Traffic Model Regression Models


0.I Response Model
Variable Regression Model Rz
0.0 Throughput S s t d = --5.072 X l O - a G a N -- 2 . 0 8 8 G a 0.99
Normalized Mean Delay Missed Deadline [ Collision Ratio
Throughput (S) (sec) (D) Ratio (F) I (C) + 2 . 6 5 8 G 2 + 0.11.63
SnT = - 1 . 4 9 6 G ~ + 2 . 1 6 5 G z + 0.143 0.99
[] IEEE ,802.11 0.6421 10.8162 [ 0.5926 [ 0.2079 [
ia rc only 0.6868 o.oa,a o.o,ez [ 0.07z4 Delay (sec) lgst d : ( - 0 . 4 4 4 G a N + O,763GZN 0.97
-0.393GN + 7.192G 3 - 11.184G 2
[=ECA Only [ 0:6948 , 0:0410 i 0"0079 t 0.0096
+ 5 . 4 9 7 G + 0 . 0 6 1 N - 0 . 0 4 3 ) 2°
121RT-MAC ] 0.691_8
........ 0,0383 _ 0,0068 ~ 0.0091 .......
DRT : 1.239G a - 1.699G z + 0.778G 0.99
+ 4 . 4 5 9 x 1 0 - 3 E - 0.103
Figure 7: RT-MAC Components Study Missed /;'6't d : sin2( -0.398Ga N + 0.511GUN 0.95
Deadline - - 0 A 3 6 G N + 5 . 7 2 6 G a -- 4 . 6 6 8 G
Ratio +1.415)
/7'~T = 8.6t6 X [()-aGUN + 1.810G a 0.98
of F and C, RT-MAC was most effective in minimizing these - 2 . 6 6 6 G 2 + 1.235G - 0.180
Collision Cstd : sinU( - 0 - 3 7 6 G a N + 0 . 6 6 1 G 2 N 0.97
metrics. In the case of D, RT-MAC was higher than TC Only
Ratio - 0 . 3 4 1 G N + 0 . 3 5 7 G a -F 0 . 0 5 3 5 N
demonstrating that the CW value (CW= 8N) is not optimal +0.0336)
with respect to mean delay. CnT = 5,995 × 1 0 - z G a - 2.626 × i O - Z G z 0.97
-5.150 × 10-8N 3
Similarly, to determine the relative contribution of CW ex- + 1 . 5 2 7 × 1 0 - 4 N + 2.013 × 1.0 - 4
pansion alone versus CW expansion and tracking BVs we ran
simulations using a network with an ideal channel, telemetry
traffic model, and N = 10, 50; G = 0.7. We found that C regression model. In general, the larger the R 2 value, the better
was reduced the most when both aspects of the collision avoid- the model fits the observed data (l.0 maximum). All factors in
ance algorithm were active. Both the CW expansion and BV the regression models and their interactions were shown to be
tracking reduced collisions roughly equally when enabled sep- significant to probability levels less than 0.0001 and, therefore,
arately. should be retained. One of the assumptions used in a regres-
sion is that the errors (residuals) are normally distributed [17].
V.B. R e g r e s s i o n Models Residuals are the difference between the regression model pre-
diction and the simulation data. To verify normality, when the
Response variable data was gathered for all combinations of
residuals are plotted in a normal quantile-quantile graph, the
the factors from Table 3 for both traffic models. The statistical
resulting line should be linear. The normal quantile-quantile
significance of the data was then evaluated using the SAS sta-
graphs for the regression models residuals in Tables 6 and 7
tistical package [30]. The general linear model (GLM) proce-
were generally quite linear indicating that, in fact, the residu-
dure was used to evaluate the impact of the simulation factors
als were normally distributed.
and their interactions with the response variables. Data trans-
Tables 6 and 7 contain the regression models for the teleme-
formations were required for DStd, Fstd, and Cstd due to the
wide range of values obtained. After trying several transfor- try traffic model and the avionics traffic model, respectively.
mations, it was determined that the power transformation (with These models are valid for 5 _< N < 50, 0.3 _< G _< 0.9 and
a = 0.05) resulted in the best regression model for Dstd and E = 0, 1 where N is the number of stations in the network, G
the arcsin transformation was best for Fsta and Cstd. is the total offered load, and E = 0 means an ideal channel and
E = 1 means an errored channel as described in Section IV.
For each response variable in both traffic models, a regres-
sion model was developed and an R 2 value was calculated. In Section V.A we stated that RT-MAC stabilized the be-
Statistically, R 2 measures how much variation in the response havior of the response variables. To support this claim we
variable can be accounted for by the factors used to develop the note that RT-MAC, in general, requires fewer terms and

Mobile Computing and Communications Review, Volume 3, Number 2 25


those terms have fewer GZN v interactions than the corre- [2] N. Abramson. The throughput of packet broadcast-
sponding IEEE 802.11 models. Specifically, we observe that ing channels. IEEE Transactions on Communications,
F n r (telemetry traffic model) and SR:c, DnT' (avionics traf- COM-25(1): 117-128, January 1977.
fic model) are virtually independent of N. in contrast to
IEEE 802.11, RT-MAC provides both better performance and [3] ASC/ENA Wright-Patterson AFB, OH, Wright-
a graceful degradation of performance in high network de- Patterson AFB, OH. Digital Time Division Co,z-
mand situations. mand/Response Multiplex Data Bus, MIL-STD- 1553B
The simulation factor E is conspicuous in the regression edition, September 1978.
models by its absence. Indeed, it only appears in the DRT. [4] R. O. Baldwin, N. J. Davis IV, and S. F. Midkiff. Imple-
model for avionics traffic. It was found that, while usually mentation of an IEEE 802.1 t wireless LAN model us-
statistically significant, the effect of an errored channel was
ing OPNET. In Proceedings ofOPNETvVORK'98. MIL3,
masked by the effects of either G or JV or both. We attribute Inc., June 1998.
this to two reasons. First, the mean amount of time in which
errors can occur is quite small compared to the error-free time. [51] R Bhagwat, R Bhattacharya, A. Krishna, and S. Tripathi.
Second, especially in the case of the telemetry traffic model, Enhancing throughput over wireless LANs using channel
the amount of data that must be retransmitted when an error state dependent packet scheduling, in Proceedings 1EEE
does occur is relatively small and much of the time in the er- INFOCOM '96, pages 1133-1140. Institute of Electrical
rored state is spent waiting for an acknowledgement. By the and Electronics Engineers, 1996.
time the next transmission occurs, much of the time in the
errored state has past. However, it should not be concluded [6] R Bhagwat, R Bhattacharya, A. Krishna, and S. Tripathi.
that bit errors do not have a discernible effect on network Using channel state dependent packet scheduling to im-
performance--the simulation data indicates they do. Rather, prove TCP throughput over wireless LANs. Wireless
we conclude that for the error model employed, the regression Networks, 3:91-102, 1997.
models were influenced to a higher degree by N and G.
[7] V. Bharghavan. Performance evaluation of algorithms
for wireless medium access. In IEEE International
VI. Conclusions and Future Work Computer Pe@)rmance and Dependability Symposium.
IPDS'98, pages 86-95. Institute of Electrical and Elec-
We have presented a simple, elegant real-time enhancement to tronics Engineers, 1998.
the IEEE 802.11 medium access control (MAC) protocol, RT-
MAC. We have shown that by implementing a transmission [8] G. Bianchi, L. Fratta, and M. Oliveri. Performance evalu-
control policy and sharing station BVs that dramatic perfor- ation and enhancement of the CSMA/CA MAC protocol
mance improvements can be realized. Further, we demonstrate for 802.11 wireless LANs. The Seventh tEEE bzterna-
that network behavior in terms of mean delay, missed dead- tional Symposium on Personal, Indoor and Mobile Radio
line ratio, and collision ratio is stabilized. By using RT-MAC, Communications PIMRC '96, pages 392-396, October
packet missed deadlines and mean delay became virtually in- 1996.
dependent of the number of stations in the network. Addi-
tionally, we developed regression models that can be used to [9] C. R. Braun. Wireless LAN standard proposal highlights
predict key performance measures of both IEEE 802.11 and the need for speed. Wireless Systems Design, 3(9): 13-16,
RT-MAC networks. September 1998.
We are currently investigating the performance of RT-MAC [10] E Call, M. Conti, and E. Gregori. IEEE 802.tt wire-
in an ad hoc network with stations that are transmitting real- less LAN: Capacity analysis and protocol enhancement.
time voice traffic along with an increasing load of non-real- In INFOCOM '98, Conference on Computer Communi-
time traffic. Further, we are studying the effect on network per- cations, pages 142-149. Institute of Electrical and Elec-
tormance of adding IEEE 802.11 stations to a network consist- tronics Engineers, 1998.
ing of RT-MAC stations. We also intend to determine whether
mean delay can be further improved by optimizing the CW [11] T. Carpenter, K. Driscoll, K f H o y m e , and J. Carciofini.
expansion factor. ARINC 659 scheduling: Problem definition. In 1994
RT-MAC was evaluated using a wireless LAN. The im- Real-Time Systems Symposium, pages 165-169. Institute
provements offered by our protocol, however, are not limited of Electrical and Electronics Engineers, 1994.
to wireless LANs or even to real-time data traffic. Our re-
sults extend to any time-slotted LAN, wired or wireless. Fur- [t2] B. Crow, I. Widjaja, J. G. Kim, and R Sakai. Perfor-
ther, our enhanced collision avoidance algorithm can be im- mance of IEEE 802.11 wireless local area networks. In
plemented independent of the data being transported. Proceedings of the SPIE, volume 2917, pages 480-491.
The International Society of Optical Engineers, 1996.

References [13] R. Dube, C. Rais, and S. K. Tripathi. Improving NFS


performance over wireless links. IEEE Transactions on
[1] N. Abramson. The ALOHA system-another alternative Computers', 46(3):290-298, March 1997.
for computer communications. In AFIPS Conference
Proceedings 1970 Fall Joint Computer Conference, vol- [14] Editors of IEEE 802.11. Wireless LAN Medium Access
ume 37, pages 281-285, 1970. Control (MAC) andPhysical Layer (PHY) Specifications,

26 Mobile Computing and Communications Review, Volume 3, Number 2


Draft Sta~da~d 802.11, P802. ] ]/D6.1. Institute of Elec- [30] The SAS System. SAS Institute, Inc. Cary, NC.
trical and Electronics Engineers, Inc., New York, May
[31] F. S. Tobagi and L. Kleinrock. Packet switching in radio
1997.
channels: Part II-the hidden terminal problem in carrier
115] J. Etlsberger, D. Hogrefe, and A. Sm'ma. SDL, Pormal sense multiple-access and the busy-tone solution. IEEE
Object-Oriented Language for Communicating Systems. Transactions on Communications, COM-23(12):1417-
Prentice Hall Europe, Hertl~rdshire, UK, 1997. 1433, December 1975.

[16] L. Kleinrock F. S. Tobagi. Packet switching in radio [32] F. S. Tobagi and L. Kleinrock. Packet switching in radio
channels: Part III-polling and (dynamic) split-channel channels: Part IV-stability considerations and dynamic
reservation multiple access. IEEE Transactions on Com- control in carrier sense multiple access. IEEE Transac-
munications, COM-24(8):832-845, August 1976. tions on Communications, COM-25(10): 1103-1119, Oc-
tober 1977.
[ 17] R. Jain. The Art of Computer Systems Pelformance Anal-
ysis. John Wiley and Sons, Inc., New York, 1991. [33] E S. Tobagi and L. Kleinrock. The effect of acknowl-
edgment traffic on the capacity of packet-switched ra-
[18] L. Kleinrock. Queueing Systems Volume 1: 77~eory. John dio channels. 1EEE 7?ansactions on Communications,
Wiley and Sons Inc., New York, 1975. COM-26(6):815-826, June 1978.

[19] L. Kleinrock. Queueing Systems Volume 2. Computer [34] M. A. Visser and M. E1 Zarki. Voice and data transmis-
Applications. John Wiley and Sons, Inc., New York, sion over an 802. l 1 wireless network. In 6th IEEE In-
t976. ternational Symposium on Personal, bldoo~ and Mobile
Radio Communications, volume 2, pages 648-652. Insti-
[20] L. Kleinrock. Principles and lessons in packet comnm- tute of Electrical and Electronics Engineers, 1995.
nications. Proceedings of the IEEE, 66(11): 1320--1329,
November 1978. [35] J. Weinmiller, H. Woesner, and A. Wolisz. Analyzing and
improving the IEEE 802.1 l-MAC protocol for wireless
[21] L. Kleinrock and M. O. Schotl. Packet switching LANs. In MASCOTS 96. Proceedings of the Fourth Inter-
in radio channels: New conflict-free multiple access national Workshop on Modeling, Analysis, and Simula-
schemes. IEEE Transactions on Communications, COM- tion of Computer and Telecommunication Syswms, pages
28(7):1015-1029, July 1980. 200-206, February 1996.
[22] L. Kleinrock and E A. Tobagi. Packet switching in ra- [36] W. Zhao and K. Ramamritham. Virtual time CSMA pro-
dio channels: Part I: CSMA modes and their throughput- tocols for hard real-time communication. IEEE Transac-
delay characteristics. IEEE Transactions on Communi- tions on Software Engineering, SE-13(8):938-952, Au-
cations, COM-23(12): 1400-1416, December 1975. gust 1987.
[23] J. E Kurose, M. Schwartz, and Y. Yemini. Multiple-
access protocols and time-constrained communication. Biographies
Computing Surveys, 16(1):43-70, March 1984.
Rusty O. Baldwin is a captain in the U.S. Air Force and a
[24] T. Liu, J. A. Silvester, and A. Polydoros. Performance Ph.D. student in electrical engineering at Virginia Tech. He
evaluation of R-ALOHA in distributed packet radio net- received his B.S.E.E. degree (with honors) from New Mexico
works with hard real-time communications. In 1995 State University in 1987, and his M.S. degree in Computer
IEEE 45th Vehicular Technology Co@rence, volume 25, Engineering from the Air Force Institute of Technology in
pages 554-558. Institute of Electrical and Electronics 1992. His research interests include computer communication
Engineers, Inc., July 1995. protocols, software engineering, and wireless networking.

[25] M.H. MacDougall. Simulating CompuWr Systems: Tech- Nathaniel J. Davis IV is an Associate Professor in the
niques and Tools. The MIT Press, Cambridge, MA, Bradley Department of Electrical and Computer Engineering
1992. at Virginia ]~ch. He received his B.S. and M.S. degrees
from Virginia Tech and his Ph.D. degree from Purdue
[26] N. Malcohn. Hard Real-l~me Communications in High
University. His research interests include wireless commu-
Speed Networks. PhD thesis, Texas A&M University,
nications networks, computer performance evaluation, and
December 1994.
high-performance computer architectures.
[27] N. Malcolm and W. Zhao. Hard real-time communica-
tions in multiple-access networks. Real-7]'me Systems, Scott E Midkiff is an Associate Professor in the Bradley
8(1):35-77, January 1995. Depamnent of Electrical and Computer Engineering at
Virginia Tech. He received his B.S.E. degree in 1979 from
[28] MIL 3, Inc., 3400 International Drive, NW Washington Duke University, his M.S. degree from Stanford University
D.C., 20008. OPNETModeler, 1997. in 1980, and his Ph.D. from Duke University in 1985. His
[29] J. L. Sobrinho and A. S. Krishnakumar. Real-time traffic research interests include wireless networking, network man-
agement related to quality-of-service and capacity planning,
over the IEEE 802.11 medium access control layer. Bell
performance evaluation, and multimedia applications.
Labs Technical Journal, pages 172-187, 1996.

Mobile Computing and Communications Review, Volume 3, Number 2 27

Вам также может понравиться