Adaptive Power Saving Algorithm for Mobile Subscriber Station in ...
Adaptive Power Saving Algorithm for Mobile Subscriber Station in 802.16e Omanand Jha Vatsa, Mayank Raj, Ritesh Kumar K, Deepak Panigrahy and Debabrata Das International Institute of Information Technology - Bangalore, Electronics City, Bangalore 560 100, India. E-Mail: {omanandjha.vatsa, mayank.raj, riteshkumar.k, deepak.panigrahy, ddas}(iiitb.ac.in
Abstract - WiMAX standard for mobility (IEEE 802.16e) addresses the issues related to energy conservation at Mobile Subscriber Station (MSS) in sleep mode. In 802.16e the sleep mode is exponentially distributed according to traffic [2]. The algorithm that deals with sleep mode energy consumption in 802.16e network does not address the problem of varying traffic rates. This paper proposes an adaptive sleep mode interval
control algorithm for 802.16e, which takes into account the downlink traffic pattern to minimize energy consumption. Moreover, we characterize the standardized sleep mode in IEEE 802.16e MSS taking into account various situation like, moving
into sleep, listening or wake-up modes. It has been observed that our proposed algorithm for sleep mode saves substantial amount of energy at lower traffic and almost same at higher traffic as compared to 802.16e. The results reveal that the optimization of sleep mode leads to decrease in average waiting time delay at MSS. Furthermore, our analytical evaluations are close to simulation results. Keywords-802.16e; Power Saving Class A; Sleep mode; MSS;
WiMAX
I. INTRODUCTION The popularity of wireless access network is growing due to its major inherent property to support mobility and higher bandwidth with respect to technological advancement. Out of wieles aces he WiMAX iMA [II] eem to differentdiffren wireless access newors, networks, the [11] seems to be promising due to higher bandwidth support and Cadvent of mobility standad IEEE
802.16e.Class mobility standard IEEE 802. 16e.
The IEEE 802.16e standard provides enhancements to IEEE 802.16d [1] to support mobile subscriber stations (MSSs) moving at vehicular speeds and thereby specifies a system for combined fixed and mobile broadband wireless access. Mechanisms to support higher laye layer hadoe handover bewenbs between base stations or sectors are specified. It fills the gap between broadband wireless access networks and high mobility cellular networks. Moreover, according to present innovation it can provide higher bandwidth per user than cellular network.
Mehnsst*upr hihr-
The battery charge is finite in an MSS, hence the conservation of energy is a challenging issue. The mechanism for efficient management of limited energy is thus turning into a very significant area of research. One popular technique to save
1-4244-0614-5/07/$20.00 ©2007 IEEE.
battery charge is "sleep mode", i.e., sleep mode is a state in which an MSS conducts pre-negotiated periods of absence from the serving base station (BS) air interface. The periods are characterized by the unavailability of the MSS, as t
BS, to DL )(dwnflink:HfromeBS t observedUfromptheksrving MSS) or UL ulink: from MSS to BS) traffic. However a
few works have been reported in literature on energy conservation at MSS with respect to mobility in WiMAX 802.16e [2, 3, 4, 5, 7, 8, 9].
In [3], the authors have addressed issues like analytical modeling of energy consumption in sleep mode due to downlink traffic and later extended to cases involving uplink traffic as well [5,10]. Issues of cell size and effect of handover
on sleep mode havebeen dealt [4] as well. Though, significant performance improvement has been reported in the proposed algorithm in [4], but issues such as adapting the sleep intervals as per traffic rate have not been addressed effectively. o
The 802.16e [2] currently addresses the classification of connections having common demand properties into one of the threethe Power Saving Classes. Power Class A is Sandaroest Eor (Eand Non realTme- Varabe Rate (NR t contions. Pow Cla BBisiS recommended UNsTed GranetSice .recom endefor Unsolicited Class Grant Service (UGS) adRa ieVral ae(TV)cnetos (UGS) oe connections. ass C iSisefor iast and mangeen for multicast and management connections. Algorithm of choosing Power Saving Class type for certain connections has not been defined in the standard. In Power Saving Class A sleep interval of standard 802.16e, the MSS negotiates minimum sleep period when it has te conneo S imumsleep no active connection with BS. However, this minimum sleep period increases increases by truncated-binary exponential backoff p ack of algorithm [2] till t if there are no packets to transmit or receive, where, t>MX is the minimum allowed sleep interval and t1 is the maximum allowed sleep interval. The major issue in case of Class A is, once tmpz is defined it remains fixed, as a result of which sometimes MSS stays in larger sleep intervals even though traffic is waiting for transmission or sometimes has smaller sleep intervals, which results more consumption of battery power in sleep mode due to frequent n
period
w
v
ibtredary
listening intervals. To the best of authors' knowledge, this paper has proposed a novel adaptive algorithm which has optimized the sleep mode with respect to traffic and decreased the average delay for packet reception with respect to 802.16e. Moreover, the proposed algorithm has been presented by analysis and supported by simulation. It has been observed that the analytical and simulation results are very close. The rest of the paper has been organized as follows. Section II presents the proposed algorithm for dynamic sleep interval. Section III presents the analytical model for the dynamic sleep interval algorithm. Section IV presents the simulation techniques to evaluate the algorithm, and discusses part of the simulation and the results in comparison to standard. The paper ends with concluding remarks in Section V. II. PROPOSED ALGORITHM FOR SLEEP MODE OPERATION The IEEE 802.16e standard gives the algorithm for determining the sleep interval in Power Saving Class A as,
approaches tm as shown in Eq. (1). In our proposed algorithm we intend to adapt t,1F based on DL traffic pattern to predict the next time of arrival of the DL frame. This will decrease the number of listening intervals in the sleep mode, thus energy consumed in sleep mode. As the sleep interval approaches ; we icrease the sleep interval incrementally as an average of n1th sleep interval and , as in Eqns (25), thus decreasing the delay DL frame has to incur waiting for the MSS to wake up. In the Section III, an analytical model for energy consumption and delay has been presented with respect to the sleep mode evaluated from proposed dynamic algorithm. III.
ANALYSIS OF SLEEP MODE EFFECT ON ENGERGY AND DELAY
Wake up Sleep Mode Wake up Slee
C1tF trnax J2-1trn, ~2eother(1)i/e
C_
~tyn
x
The probability that a DL frame arrives at BS for an MSS in is given the n : sleep interval, where: _ - L :. as as
Arrival of DL traffic at BS for MSS Listening Interval (L)
A.e-
EL
(15)
frame available for transmission at MSS Figurea 4: ULV ..........interval J after the nth sleep interval is
Here,
L < X < La We define a new random variable t, as, W,1 < t, C w ft- , + tn = .i otherwise w-
in
the
listening
-
(20)
The average value of t (tLv t) can be calculated as r
ii_ 4AUs-Xtex d t_fs -i2`
(wiz+
eXL-
4 -.3AX,, 4
_
1. 21
,
=
> w,n + tn) P(t ,,_ (21)
- 1)L] EL +J tj,t!-aEt tt)-I
El
(22)
.-
e
tu < w;j.
(e2RuL-
'
z
1)
L+AL eL
(23)
)
Abstract - WiMAX standard for mobility (IEEE 802.16e) addresses the issues related to energy conservation at Mobile Subscriber Station (MSS) in sleep mode. In 802.16e the sleep mode is exponentially distributed according to traffic [2]. The algorithm that deals with sleep mode energy consumption in 802.16e network does not address the problem of varying traffic rates. This paper proposes an adaptive sleep mode interval
control algorithm for 802.16e, which takes into account the downlink traffic pattern to minimize energy consumption. Moreover, we characterize the standardized sleep mode in IEEE 802.16e MSS taking into account various situation like, moving
into sleep, listening or wake-up modes. It has been observed that our proposed algorithm for sleep mode saves substantial amount of energy at lower traffic and almost same at higher traffic as compared to 802.16e. The results reveal that the optimization of sleep mode leads to decrease in average waiting time delay at MSS. Furthermore, our analytical evaluations are close to simulation results. Keywords-802.16e; Power Saving Class A; Sleep mode; MSS;
WiMAX
I. INTRODUCTION The popularity of wireless access network is growing due to its major inherent property to support mobility and higher bandwidth with respect to technological advancement. Out of wieles aces he WiMAX iMA [II] eem to differentdiffren wireless access newors, networks, the [11] seems to be promising due to higher bandwidth support and Cadvent of mobility standad IEEE
802.16e.Class mobility standard IEEE 802. 16e.
The IEEE 802.16e standard provides enhancements to IEEE 802.16d [1] to support mobile subscriber stations (MSSs) moving at vehicular speeds and thereby specifies a system for combined fixed and mobile broadband wireless access. Mechanisms to support higher laye layer hadoe handover bewenbs between base stations or sectors are specified. It fills the gap between broadband wireless access networks and high mobility cellular networks. Moreover, according to present innovation it can provide higher bandwidth per user than cellular network.
Mehnsst*upr hihr-
The battery charge is finite in an MSS, hence the conservation of energy is a challenging issue. The mechanism for efficient management of limited energy is thus turning into a very significant area of research. One popular technique to save
1-4244-0614-5/07/$20.00 ©2007 IEEE.
battery charge is "sleep mode", i.e., sleep mode is a state in which an MSS conducts pre-negotiated periods of absence from the serving base station (BS) air interface. The periods are characterized by the unavailability of the MSS, as t
BS, to DL )(dwnflink:HfromeBS t observedUfromptheksrving MSS) or UL ulink: from MSS to BS) traffic. However a
few works have been reported in literature on energy conservation at MSS with respect to mobility in WiMAX 802.16e [2, 3, 4, 5, 7, 8, 9].
In [3], the authors have addressed issues like analytical modeling of energy consumption in sleep mode due to downlink traffic and later extended to cases involving uplink traffic as well [5,10]. Issues of cell size and effect of handover
on sleep mode havebeen dealt [4] as well. Though, significant performance improvement has been reported in the proposed algorithm in [4], but issues such as adapting the sleep intervals as per traffic rate have not been addressed effectively. o
The 802.16e [2] currently addresses the classification of connections having common demand properties into one of the threethe Power Saving Classes. Power Class A is Sandaroest Eor (Eand Non realTme- Varabe Rate (NR t contions. Pow Cla BBisiS recommended UNsTed GranetSice .recom endefor Unsolicited Class Grant Service (UGS) adRa ieVral ae(TV)cnetos (UGS) oe connections. ass C iSisefor iast and mangeen for multicast and management connections. Algorithm of choosing Power Saving Class type for certain connections has not been defined in the standard. In Power Saving Class A sleep interval of standard 802.16e, the MSS negotiates minimum sleep period when it has te conneo S imumsleep no active connection with BS. However, this minimum sleep period increases increases by truncated-binary exponential backoff p ack of algorithm [2] till t if there are no packets to transmit or receive, where, t>MX is the minimum allowed sleep interval and t1 is the maximum allowed sleep interval. The major issue in case of Class A is, once tmpz is defined it remains fixed, as a result of which sometimes MSS stays in larger sleep intervals even though traffic is waiting for transmission or sometimes has smaller sleep intervals, which results more consumption of battery power in sleep mode due to frequent n
period
w
v
ibtredary
listening intervals. To the best of authors' knowledge, this paper has proposed a novel adaptive algorithm which has optimized the sleep mode with respect to traffic and decreased the average delay for packet reception with respect to 802.16e. Moreover, the proposed algorithm has been presented by analysis and supported by simulation. It has been observed that the analytical and simulation results are very close. The rest of the paper has been organized as follows. Section II presents the proposed algorithm for dynamic sleep interval. Section III presents the analytical model for the dynamic sleep interval algorithm. Section IV presents the simulation techniques to evaluate the algorithm, and discusses part of the simulation and the results in comparison to standard. The paper ends with concluding remarks in Section V. II. PROPOSED ALGORITHM FOR SLEEP MODE OPERATION The IEEE 802.16e standard gives the algorithm for determining the sleep interval in Power Saving Class A as,
approaches tm as shown in Eq. (1). In our proposed algorithm we intend to adapt t,1F based on DL traffic pattern to predict the next time of arrival of the DL frame. This will decrease the number of listening intervals in the sleep mode, thus energy consumed in sleep mode. As the sleep interval approaches ; we icrease the sleep interval incrementally as an average of n1th sleep interval and , as in Eqns (25), thus decreasing the delay DL frame has to incur waiting for the MSS to wake up. In the Section III, an analytical model for energy consumption and delay has been presented with respect to the sleep mode evaluated from proposed dynamic algorithm. III.
ANALYSIS OF SLEEP MODE EFFECT ON ENGERGY AND DELAY
Wake up Sleep Mode Wake up Slee
C1tF trnax J2-1trn, ~2eother(1)i/e
C_
~tyn
x
The probability that a DL frame arrives at BS for an MSS in is given the n : sleep interval, where: _ - L :. as as
Arrival of DL traffic at BS for MSS Listening Interval (L)
A.e-
EL
(15)
frame available for transmission at MSS Figurea 4: ULV ..........interval J after the nth sleep interval is
Here,
L < X < La We define a new random variable t, as, W,1 < t, C w ft- , + tn = .i otherwise w-
in
the
listening
-
(20)
The average value of t (tLv t) can be calculated as r
ii_ 4AUs-Xtex d t_fs -i2`
(wiz+
eXL-
4 -.3AX,, 4
_
1. 21
,
=
> w,n + tn) P(t ,,_ (21)
- 1)L] EL +J tj,t!-aEt tt)-I
El
(22)
.-
e
tu < w;j.
(e2RuL-
'
z
1)
L+AL eL
(23)
)
