Select topics on telecommunications

Packet-aware Transport for Metro Networks - in progress

Literature Survey

1. Current transport networks - packet switching equipment is consolidated at few locations, hence packet data is not switched as close as possible to the end users as desired. Particularly, Ethernet transport is not efficient.

Virtual Private Networks (VPNs) - Layer 2 and Layer 3 VPNs - 06/07/2004

Literature Survey for VPNs

1. L2 VPNs - point to point pseudo-wire service, and point-to-multipoint LAN emulation service
2. Issues in L2 VPNs - provisioning of pseudo-wire services, preventing routing loops in virtual private LAN services, interworking between different L2 technologies at endpoints of the VPN
3. L3 VPNs - classification into BGP/MPLS IP VPNs, virtual router IP VPNs (both PE-based) and CE-based IPSec VPNs.

Virtual Private Networks (VPNs) - Architecture, Scenarios, Requirements for Standardisation - 06/03/2004

Literature Survey for VPNs

1. Comparison of CE and PE-based VPNs
2. Differentiation between VPNs based on the service offered - L3, L2 and L1 VPNs
3. VPN architecture issues - endpoint creation, endpoint discovery, tunnel establishment and distribution of reachability information.

Generic Framing Procedure - 07/30/2004

Literature Survey for GFP

1. Supports variable-length payload and hence there is no need for SAR functions and frame padding.
2. Since there is no transparency processing, the unpredictable bandwidth expansion common in PoS and HDLC solutions is not present in GPF.
3. GPF eventually should replace PoS solutions. It also works well with RPR (how?)

IEEE 802.17 RPR (Resilient Packet Ring) - 05/07/2004

Literature Survey for IEEE 802.17 RPR

Questions

1. RPR is an upcoming standard for a ring-based MAN/WAN. Other ring-based network architectures include Token Ring (LAN), FDDI (LAN/MAN) and ATM-R (LAN/MAN).
2. Present-day MAN/WAN transport is typically SONET/SDH rings. SONET/SDH rings are characterised by static bandwidth allocation to the various nodes on the ring. Hence, they are not efficient for applications such as bridging two Ethernets at different locations.
3. In Token Ring, the frame goes around the ring back to the sender, thus serving as an acknowledgement to the sender. But in RPR, the receiving station removes the frame from the ring, thus allowing the simultaneous re-use of the ring between two other stations not involved in this particular transmission. This re-use is termed spatial reuse.
4. Ring access in RPR is using buffer insertion , unlike Token Ring, where the station with the token has access to the ring. Every station has a buffer called the transit queue , in which frames in transit must be queued temporarily.
5. A station transmits only if there are no frames in transit, and the transit queue is empty.
6. A frame that arrives in the transit queue when the station is transmitting a frame, has to wait until the transmission is complete.
7. RPR design allows for a station to have a single transit queue, or two queues, a primary and secondary.
8. RPR implements a fairness algorithm to share ring bandwidth among various stations. The algorithm involves the downstream station sending control frames to the upstream stations to reduce transmit rates to a fair rate.
9. Each station maintains a topology database which is updated by periodic topology frames received from all the stations in the ring. Topology frames are also sent when there is a change in the ring such as the addition/removal of a station, or the failure of a link/station.
10. Resilience is provided in RPR by steering, which means the station that detects a failure sends a new frame in the only available ring.
11. RPR supports transparent bridging using the extended frame format, which can encapsulate other IEEE 802 frames.

MPLS - 05/05/2004

Literature Survey for MPLS

Questions

1. Conventional IP routing requires every router in the path to re-examine the IP header to determine the next hop.
2. In MPLS, ingress router determines the flow (Forwarding Equivalence Class or FEC) for this packet, and adds a label to the IP header, and sends it to next hop. The next router looks at the label, identifies the flow, and sends it on.
3. Label is an index into a table that specifies the next hop, and the new label, which the next hop router needs to identify the flow. The label has only local significance, ie. between one label-switching router and the next.
4. MPLS makes 'source routing', i.e. forcing a particular path to be used for a packet, possible. This routing might be a matter of policy or for traffic shaping.
5. Other examples of label switching include switching based on VPI/VCI in ATM, DLCI in frame relay. So is MPLS label-switching also L2 switching?
6. If ATM is the L2 protocol underneath the IP network, then the label will be carried in the ATM VCI/VPI field. This also allows VC/VP merging. Similarly, if Frame Relay is the underlying transport, the label is carried in the DLCI section of the FR header.

GSM Mobility Management and Issues