320 likes | 460 Views
Wireless MAN IEEE 802.16. By Sameer Kurkure KReSIT IIT Bombay. Guide Prof. Anirudha Sahoo KReSIT IIT Bombay. Overview. Broadly applicable to the systems operating between 11-66Ghz Point to Multipoint Metropolitan Area Network Connection oriented Support difficult user environment
E N D
Wireless MAN IEEE 802.16 By Sameer Kurkure KReSIT IIT Bombay Guide Prof. Anirudha Sahoo KReSIT IIT Bombay
Overview • Broadly applicable to the systems operating between 11-66Ghz • Point to Multipoint • Metropolitan Area Network • Connection oriented • Support difficult user environment • High bandwidth, large number of end users • Continuous and burst traffic • Efficient user of spectrum • Flexibility in offering QoS for • CBR, rt-VBR, nrt-VBR, and BE according to their classes • Supports multiple 802.16 PHYs • LOS communication
Basic Architecture • Single BS (Base Station) • Multiple fixed SS (Subscriber Stations) associated with one BS • Hundreds of users in one SS
MAC Mechanism and Framing • BS periodically grants transmission opportunities • Grants are time slots on uplink channel • Some time slots reserved for contention • Broadcasting MAP message • Downlink subframe contains • DL-MAP • UL-MAP
Scheduling Services • Unsolicited Grant Service (UGS) • Fixed size data on periodic basis e.g. VoIP • Bandwidth is separately reserved • Real Time Polling Service (rtPS) • Variable size data on periodic basis e.g. MPEG video • By unicast requests • Non-real Time Polling Service (nrtPS) • Variable size data e.g. FTP service • Requires some minimum bandwidth • Best Effort (BE) • Service flows like HTTP (no minimum bandwidth) • Use contention window for requests
Bandwidth Allocation and Request Mechanism • Grant per Connection (GPC) • Bandwidth allocated • Decision made at BS for each connection • BS may be overloaded • Suitable for few SSs. • Grant per Subscriber Station (GPSS) • Fixed bandwidth for each SS • Decision for connection made at SS • Less burden on BS
QoS Architectures General
QoS Architectures General • Components • Traffic Classifier • Contention Ratio Calculator (CRC) • SS’s upstream Scheduler • BS’s upstream scheduler • Contention Slot Allocator (CSA) • Contention Resolution Algorithm (CRA) • BS’s downstream scheduler
QoS Architectures General • Components • Traffic Classifier • Contention Ratio Calculator (CRC) • Ru + Rrp + Rnp + Rbe = 100% • SS’s upstream Scheduler • BS’s upstream scheduler • Contention Slot Allocator (CSA) • Contention Resolution Algorithm (CRA) • BS’s downstream scheduler
QoS Architectures General • Components • Traffic Classifier • Contention Ratio Calculator (CRC) • SS’s upstream Scheduler • Selects traffic according to the grants from BS’s upstream scheduler • BS’s upstream scheduler • Contention Slot Allocator (CSA) • Contention Resolution Algorithm (CRA) • BS’s downstream scheduler
QoS Architectures General • Components • Traffic Classifier • Contention Ratio Calculator (CRC) • SS’s upstream Scheduler • BS’s upstream scheduler • Decide schedules from the requests obtained • Contention Slot Allocator (CSA) • Contention Resolution Algorithm (CRA) • BS’s downstream scheduler
QoS Architectures General • Components • Traffic Classifier • Contention Ratio Calculator (CRC) • SS’s upstream Scheduler • BS’s upstream scheduler • Contention Slot Allocator (CSA) • Maintain proper ratio between contention slots & reservation slots • Contention Resolution Algorithm (CRA) • BS’s downstream scheduler
QoS Architectures General • Components • Traffic Classifier • Contention Ratio Calculator (CRC) • SS’s upstream Scheduler • BS’s upstream scheduler • Contention Slot Allocator (CSA) • Contention Resolution Algorithm (CRA) • Rules to resolve contentions • BS’s downstream scheduler
QoS Architectures General • Components • Traffic Classifier • Contention Ratio Calculator (CRC) • SS’s upstream Scheduler • BS’s upstream scheduler • Contention Slot Allocator (CSA) • Contention Resolution Algorithm (CRA) • BS’s downstream scheduler • Schedule back grant along with other data.
QoS Architectures Admission Control based on scheduling service characteristics • BS decides accept / reject of user’s request • Accepting may cause QoS degradation • Some portion of bandwidth ‘d’ from is given to new connection • Cannot compromise on UGS and rtPS connections • nrtPS connection best for sacrifice • Keep assigning BW for new connection until QoS parameter is safe (Lmax degradation level) Degradation Model
QoS Architectures Admission Control based on scheduling service characteristics New Connection Assign BW for New connection Yes Cut ‘d’ from ongoing nrtPS connections L = L + 1 (initially 0) bmax – L*d >= bmin Degradation Model No Reject connection
QoS Architectures Admission Control based on scheduling service characteristics • Analytical Model • Can be expressed in state diagram • s = ( nUGS, nrtPS, nnrtPS, Ln) be the valid state • Constraints of Validity • nUGS*bUGS + nrtPS*brtPS + nnrtPS*(bmax – Ln*d > bmin) <= B • nrtPS*brtPS + nnrtPS*(bmax – Ln*d > bmin) <= B - U • Ln <= Lmax • Transition • At acceptance of new request • Completion of connection
QoS Architectures Admission Control based on scheduling service characteristics State Transition Diagram
Addition of high-speed data transfer to existing cable TV systems Components CMTS (Cable Modem Termination System) CM (Cable Modem) Works on request/grant mechanism Headend sends ack / data grant at each request Contention happens when CM doesn’t receive any of them QoS Architectures DOCSIS (Data Over Cable Service Interface Specification)
Variable length upstream frame is used Need to optimize contention mini slots j Adjust j such that Throughput of contention requests become equal to # new data packets transmitted in max. frame Assigned appropriate priorities to service flows Allocate mini slots for that flow QoS Architectures DOCSIS (Data Over Cable Service Interface Specification)
QoS Architectures DOCSIS (Data Over Cable Service Interface Specification)
QoS Architectures DOCSIS (Data Over Cable Service Interface Specification) • Handling priorities • Choose set of multiplication factors, ad such that 0 < ad < 1 • where d = 0,……,K and a0+a1+ …….aK = 1 • Associate these number at the specific flows • aK to highest priority • a0 to lowest priority • Allocate j * ad number of slots to the corresponding service flow
QoS Architectures Mesh Networks IEEE 802.16d • Area divided meshes to form a network • Components • Mesh Base Station (MBS) • Mesh Subscriber Station (MSS) • MBS periodically collects • Channel information • Resource request • MBS decides on • Routing • Scheduling Strategy
QoS Architectures Mesh Networks IEEE 802.16d • Communication • Intra-mesh communication • MSSs…MSSi…MSSd • Inter-mesh communication • MSSs..MBSsMBSd..MSSd • Traffic associated • Real Time Application • Like VoIP uses UDP • Data Application • Uses TCP • QoS requirement for both are different
QoS Architectures Mesh Networks IEEE 802.16d Routing Queue Length link(i,j) always assign ni,j slots assuming schedule to be fixed For queue to be stable Outgoing >= Incoming
QoS for Real Time (UDP) Traffic Constant Bit Rate (CBR) Like VoIP Drop the packet when out of constraint E.g. delay > 10ms Let Xi be constant amount of traffic generated at nodei Let ei be the end to end dropping probability decomposed into ei,j ni,j be number slots assign at nodej for flow from nodei then following condition must get satisfied: QoS Architectures Mesh Networks IEEE 802.16d Scheduling
QoS for Real Time (UDP) Traffic Variable Bit Rate (CBR) Like video conferencing Drop the packet when out of constraint E.g. delay > 10ms Let there are J number of VBR flows at nodei Dk(i,j) be amount of data generated by jth flow in framek Let ei be the end to end dropping probability decomposed into eb & ed Find Ci such that: QoS Architectures Mesh Networks IEEE 802.16d Scheduling Ci/J can be called as eqv BW of VBR for which MBS can treat as CBR for dropping ed
QoS for TCP Traffic Fixed Allocation Scheme Need to calculate ni,j number of slots to be allocated at nodej for traffic generated at nodei Let Required to satisfy two constraint: QoS Architectures Mesh Networks IEEE 802.16d Scheduling
QoS for TCP Traffic Fixed Allocation Scheme Also is directly proportional to throughput assigned to traffic at each nodei. Therefore, QoS Architectures Mesh Networks IEEE 802.16d Scheduling
QoS for TCP Traffic Adaptive Allocation Scheme Classify the link as G and B Defer allocation of slots to links which are bad Assign counter ck(i) for number of slots for nodei compensating missed slots Assign channel if counter exceeds CLim(i) QoS Architectures Mesh Networks IEEE 802.16d Scheduling
QoS for TCP Traffic Adaptive Allocation Scheme calculate, Now Here Rth(i) and CLim(i) are design parameter QoS Architectures Mesh Networks IEEE 802.16d Scheduling