Politecnico di Milano Advanced Network Technologies Laboratory. Beyond Standard MAC Sublayer

Transcription

1 Politecnico di Milano Advanced Network Technologies Laboratory Beyond Standard MAC Sublayer

2 MAC Design Approaches o Conten&on based n Allow collisions n O2en CSMA based (SMAC, STEM, Z- MAC, GeRaF, etc.) o TDMA based n Slot Synchronous n Collision free (NAMA, TRAMA, TDMA- W, SPARE- MAC, etc.) o Hybrid Some type of activity periods scheduling to spare energy

3 Energy ConsumpBon o Overhearing n hearing someone else s transmissions o Idle Listening n radio in recep&on mode but nothing fetched o Collision n unsuccessfull transmission o Control Overhead n transmission (recep&on) of signalling packets

4 S- MAC: Sensor MAC o Problem: Idle Listening consumes significant energy o Solu&on: Periodic listen and sleep listen Sleep listen Sleep o During sleeping, radio is turned off o Reduce duty cycle to ~ 10% (Listen for 120ms and sleep for 2s) Latency Energy W. Ye, et. al., Medium Access Control with Coordinated Adaptive Sleeping for Wireless Sensor Networks, IEEE/ACM Trans. on Networking, June 2004.

5 S- MAC: RaBonale o Each node goes into periodic sleep mode during which it switches the radio off and sets a &mer to awake later o When the &mer expires it wakes up and listens to see if any other node wants to talk to it o The dura&on of the sleep and listen cycles are applica&on dependent and they are set the same for all nodes o Requires a periodic synchroniza&on among nodes to take care of any type of clock dri2

6 Periodic Sleep and Listen o All nodes are free to choose their own listen/ sleep schedules. o To reduce control overhead, only neighboring nodes are synchronized together. o They listen at the same &me and go to sleep at the same &me.

7 SynchronizaBon o SYNC packets are exchanged periodically to maintain schedule synchroniza&on. SYNC PACKET Sender Node ID Next Sleep Time o SYNCHRONIZATION PERIOD: Period for a node to send a SYNC packet. o Receivers will adjust their &mer counters immediately a2er they receive the SYNC packet

8 PERIODIC LISTEN AND SLEEP

9 Maintaining SynchronizaBon Listen interval is divided into two parts: one for receiving SYNC packets and other for receiving RTS (Request To Send)

10 Choosing and Maintaining Schedules o Each node maintains a schedule table that stores schedules of all its known neighbors o For ini&al schedule, DO: n A node first listens to the medium for a certain amount of &me (at least the synchroniza&on period) n If it does not hear a schedule from another node, it randomly chooses a schedule and broadcasts its schedule with a SYNC packet immediately n This node is called a Synchronizer n If a node receives a schedule from a neighbor before choosing its own schedule, it just follows this neighbor s schedule, i.e. becomes a Follower and it waits for a random delay and broadcasts its schedule

11 Coordinated Sleeping o In a large network, we cannot guarantee that all nodes follow the same schedule. o The node on the border will follow both schedules. o When it broadcasts a packet, it needs to do it twice, first for nodes on schedule 1 and then for those on schedule 2. Schedule 1 Schedule 2

12 Collision Avoidance o S- MAC is based on conten&on, i.e., if mul&ple neighbors want to talk to a node at the same &me, they will try to send when the node starts listening. n Similar to IEEE802.11, i.e. use RTS/CTS mechanism to address the hidden terminal problem n Perform carrier sense before ini&a&ng a transmission

13 Virtual Carrier Sensing DIFS source RTS SIFS SIFS DATA SIFS destination CTS ACK overhearing stations NAV (RTS) NAV (CTS) Random Backoff Control packets to probe the channel RTS, CTS contain the TX dura&on Overhearing sta&ons refrain from accessing the channel by secng the NAV (Network Alloca&on Vector) RTS (Request To Send) CTS (Clear To Send) 13

14 The Virtual Carrier Sensing RTS C CTS CTS Terminal C refrains from transmitting for all the duration of the communication A-B 14

15 What do You Think of S- MAC? o Open Discussion

16 SPARE MAC Overview o Dynamic TDMA structure featuring the concept of Recep6on Schedule (RS): n Set of &me slots during which a sensor periodically ac&vates for RECEIVING data o SPARE MAC philosophy: n Each sensor is assigned a RS n Each sensors knows the RS of all its poten&al receivers I ll be available for receiving at slot k at slot k

17 SPARE MAC OperaBon & ProperBes o Fully distributed and coordinated algorithms to: n Assign RS n Spread the assignment informa&on throughout 1- hop neighbors o Each sensors theore&cally ac&vates only when: n It has traffic to send and/or receive (limited idle listening) n It actually receives the traffic des&ned to itself only (limited overhearing) n In the WUS

18 SPARE MAC framing o Slot & Frame Synchronous 1 Signalling SubFrame (SSU) Data SubFrame (DSU) N WUS M o N: # of Signaling Slots (SS) within the SSU o M: # of Data Slots (DS) within the DSU o WUS: Wake Up Slot 1 Rowe et al., RT-Link: A Time-Synchronized Link Protocol for Energy Constrained Multi-Hop Wireless Networks, IEEE SECON 2006

19 Wake Up Reliable ReservaBon ALOHA (WRR- ALOHA) o GOAL: to distribute the RS assignment throughout 1 Hop neighbors o Relies upon the transmission of reliable Broadcast Signalling Packets (BSPs) within the SSU BSP Format ID 1 N 1 M o o Frame Information (FI) Reception Schedule Information (RSI) FI specifies the status of the N slots in the SSU as observed by the sensor o BUSY correct transmission o FREE no transmission or collision RSI specifies the o The RS chosen by the sensor o The RSs of all the 1 hop neighbors of the sensor

20 WRR- ALOHA: BSP reservabon and ACKs o A Signalling Slot is RESERVED if at least one FI says BUSY AVAILABLE otherwise R A A R A R Channel Status Built up by S1 S3 S3 S2 S2 FI from S3 FI from S2 S3 S2 S1 S4 S4 FI from S4 S3 S1 S2 S4 FI from S1 o AVAILABLE slots can be used for access o The transmission is successful if the slot is coded as BUSY with the same sta&on ID in all the received FI

21 WRR- ALOHA: Energy Efficiency o BSP transmission is triggered by the ac&vity perceived in the WUS o All the sensors are ac&ve in the WUS Node 3 Wakes Up Nodes go to sleep W 1 2 W W SSU DSU

22 RS Scheduling and Data Transfer o RS Assignment Solu&ons n 1 HOP AWARE: RS is chosen not to overlap with RS of 1 hop neighbors o Collision Recovery Algorithms (CRAs) n Binary Exponen&al Back- off: o access &me (in frames) a2er i consecu&ve collisions is = rand [1,2 k 10 2 i ] if i 10 otherwise

23 SPARE MAC Dimensioning o Performances depends on: n Frame Length n Traffic dynamics (collisions) n CRAs o Design should account for: n Energy consump&on n Delivery delays n Applica&on requirements

24 Maximum Achievable Data rate o The maximum achievable data rate (R max ) is given by: BRS R (1) max= NT + MT + T o Being: SS DS WUS n B RS the payload length of the RS [bit] n T SS T DS and T WUS the dura&on of SS, DS and WUS respec&vely [s] n N the number of slot within the SSU n M the number of slots within the DSU o Problem: how to set N and M?

25 Dimensioning Guidelines o N N min n N min depends on the topology of the WSN (all sensors within a two hop cluster must have a unique BSP) o Once N is fixed, e.g.: cluster clouds,n min =7 M M min M max where n M max depends on the data rate required by the specific applica&on (1) n M min depends on the specific topology (a sensors must not have the same RS as any other sensor within the same one hop cluster) e.g.: Grid, M min = 2

26 R max vs M Bandwidth SS length DS length WUS length 250 kb/s 50 byte 560 byte 9 byte o Maximum RS data rate decreases with M and N

27 Energy Analysis (1) o o o It is worth having quan&ta&ve means to determine the energy efficiency Assump&ons: 1. In the generic slot i a sensor can be: o Idle (id) o Receiving (rx) o Transmicng (tx) o Sleeping (sl) 2. Traffic generated by a specific sensor toward each receiver is Poisson distributed with parameters G t 3. Overall received traffic by a specific sensor is Poisson deistributed with parameter and G r 4. RS =1 From assump&on 2 and 3, the total offered traffic in transmission (G t* ) and recep&on (G r* ) is given by: G * r = e G G G G r r * t * t * G * G r 1 G r e t 1 = G t G * t

28 Energy Analysis (2) o The power consumed within a frame depends whether the sensor is: transmicng and/or receiving or idle o The probability for a sensor to be transmicng towards i out of R recipients in DSU is: p tx (i) p = rx R * * G t G t (N i) i (0) = e ( 1 e G ) e 0 i * * r G r p rx (1) = 1- e R, R M 1 o The probability for a sensor to be receiving from j є {0.1} in DSU is: o The joint probability of transmitting to i recipients and receiving from j transmitter in DSU is: p tx(i),rx(j) = p (i) p (j) tx rx 0 i R, j {0,1}

29 Energy Analysis (3) o o o o The average consumed energy in a frame can be es&mated as: E f j= 0 i= 0 being E(i; j) the energy consumed in a frame when transmicng to i recipients and receiving from j transmirers E(i, j) = e x y is the energy consumed in state y є {rx, tx, id, sl} in a slot of type x є {signalling (s), data (d), wake- up (w)} The average consumed power per frame is: = i e rx s 1 + R j e p tx s + (M i P = f E + f T frame tx(i),rx(j) (N i j)e sl d + i e E(i, j)e tx d sl s + j) + e j e id w rx d +

30 SimulaBon Analysis o Proofs with ns2.29 in fully connected WSNs (a) and covergecast WSNs (b) a) b) o The measured confidence index for all collected sta&s&cs is below 5% in 98% of all cases PARAMETER Simulation Run Length Bandwidth Packet Length SS length DS length WUS length Ptx Prx, Pid Psleep VALUE 1000 [s] 250 [kb/s] 512 [byte] 50 [byte] 560 [byte] 9 [byte] 24 [mw] 13.5 [mw] 5 [uw]

31 Power Model o Consumed power decreases with M

32 SPARE MAC vs SMAC o Objec&ve: spare energy o SPARE MAC outperforms SMAC in terms of consumed power

33 What do You Think of SPARE- MAC? o Open Discussion