Mobile broadband problems (dealing with IP)

Mobile broadband have been a great tool for customers but often problems occur.  Most mobile networks use a hybrid of TDM and IP.  AS legacy TDM networks, such as Dignet from Telkom, are migrated to fibre the end point equipment in the tower is not changed.  Instead the TDM network is encapsulated over IP and transmitted over fibre.

A well known STP vendor involved is Huawei.  Some interesting issues are highlighted in this url: http://www.huawei.com/enapp/198/hw-080256.htm
 

IP-based signaling

Five issues on IP-based signaling

It is acknowledged industry-wide that the following issues require carriers' attention prior to constructing IP-based signaling networks:


  1. Independent signaling network
    In a way that is unrelated to circuit connections, MAP and CAP signaling dominate mobile networks and the two are generally employed for mobile user location updates, route queries and intelligent network service operations. In terms of protocol stacks, whether MAP and CAP signaling is carried via TDM, ATM or IP protocol, the SCCP layer is always necessary. That is, MAP signaling and CAP signaling need the transmission function of the SCCP layer to transmit messages to destinations. In each 2G network, a major function of STP in the signaling network is GT translation and message transfer. In each 3GPP R4 network, the SCCP-based addressing mode remains unchanged, so the GT translation and message transfer functions remain essential.
    Large-scale R4 networks demand independent STP equipment for MAP/CAP signaling bearing and transfer so as to reduce the number of signaling link sets and SCTP associations in the MSC/MSC Server. Future value-added services can therefore be flexibly supported. Multiple network elements are directly connected to a pair of STPs with large capacity, high performance in order to reduce transmission links, simplify network structure, reduce OPEX and ensure network expandability. Consequently, the hierarchical architecture is applicable to TDM, TDM-to-IP evolution and IP-based signaling networks. Carriers can choose to adopt a two or three-layered hierarchical architecture according to actual network condition.
  2. Link-based protocols
    SIGTRAN protocol stack includes the M3UA, M2PA, M2UA, SUA, and SCTP protocols. Different protocols can be selected according to link types when IP bearing is adopted to facilitate smooth upgrades for signaling networks.
    M3UA is a 3GPP recommended SIGTRAN protocol that is adopted between terminal offices and the STP. It supports all existing mobile network protocols including BICC, ISUP, MAP and CAP, but it describes an implementation protocol mainly designed for SG applications. It is limited to being a one-hop signaling transfer or as an IP SP edge access protocol.
    M2PA-based IP signaling links provide link-based signaling network management functions via MTP3. M2PA can be used as the SIGTRAN protocol between STPs that supports signaling link fault switching and enables competent network management and security.
  3. Reliability processing
    If either IP or TDM are used to bear signaling, then only the bearer layer and application protocols can be changed. Most signaling networks currently adopt dual-plane networking, and the same layer equipment is interconnected by at least two links. This reliability processing mechanism plays an important networking function, and the reliability of each signaling network can be further heightened by employing IP bearer network and SIGTRAN protocol stack features.
    The IP bearer network adopts some redundancy design features such as dual uplinks and planes. VRRP, VGMP and HRP technologies can be employed to execute switching redundant structure switching, and APDP/BFD technologies can be utilized for fault detection.
    In view of equipments, SCTP's multi-home feature can be used to improve reliability. This feature means for the same association, multiple IP addresses may be adopted at both the local and peer ends and guarantees layer-2 transmission reliability. Based on this feature, paths can be configured in different physical network segments, thus ensuring the reliability of IP bearer network links.
  4. Solving error codes
    Due to the characteristics of the IP bearer network and signaling services' transmission requirements, QoS remains an important factor underpinning SS7 signaling networks. In the TDM-based bearing mode, signaling messages are transmitted via dedicated networks. Therefore, the element affecting the bearer network QoS is error codes. An error code detection mechanism is defined into the MTP2 layer that can quickly detect error codes and implement message re-transmission and link switching to minimize losses. However, as the IP bearer network is a best-effort network, it is incapable of collecting delay, packet loss and jitter data. The SCTP protocol cannot, therefore, define physical layer requirements or quickly realize message retransmission and link switching. For each signaling network, high QoS must be guaranteed to offer users a holistic, high-grade service experience.
    Similar as the above analysis, the QoS of SS7 signaling network can be guaranteed from the network aspect and the equipment aspect. Each bearer network must enable dedicated usage, light bearing and quality guarantee. Via quality guarantee in the IP bearer network, QoS needed in the service layer can be enabled. For example, MPLS VPN or light-bearing IP dedicated network can be used to avoid network congestions.
    As equipments at both ends of IP links are on the transmission layer, the SCTP protocol can be used to improve QoS. The connection-oriented SCTP can ensure security and correctness in message transmission; it can also well control packet losses. The SCTP protocol can control and measure channel delays, which can meet the requirements for delay on the application layer. Equipment QoS can be improved via the flow control function on the service layer.
    MSC server can analyze signaling messages according to service types and customer types. When it detects poor IP link quality, it preferentially guarantees the transmission of signaling messages of important services from important customers. The STP equipment in the signaling network can preferentially transmit important signaling messages according to carriers' requirements and message contents, including origination signaling point code, destination signaling point code, and message types.
    In addition to QoS guarantees, the equipment lends itself to quantifying QoS, allowing carriers instant problem detection and an immediate awareness of signaling bearer quality. Just as the SCTP ensures reliable message transmission, the signaling transmission status of the SCTP can be used to judge the IP bearer network's QoS. An inferior QoS can increase SCTP message acknowledgement delays and the need for message retransmissions. By analyzing the two parameters - SCTP message acknowledgement delays and the number of message retransmission - the IP bearer network's QoS can be judged.
  5. Ensuring broadband link performance
    TDM-based SS7 signaling links are restricted by physical transmission rates. An ordinary link is 64Kbit/s, or 56Kbit/s in some networks, while high-speed rates are 2Mbit/s. When an IP-based SIGTRAN link is adopted, the theoretical bandwidth between the two ends can reach hundreds or thousands of megabits. However, given that the IP network is a best-effort network and that the bearer layer shares its bandwidth, the basis for ensuring broadband link performance rests with guaranteeing sufficient signaling message QoS. In this context, broadband link reference loads define another networking precondition.

So it is my theory that insufficient QoS on mobile broadband networks causes a significant amount of problems.

The factors described in this article all affect throughput as a factor of distance from the tower but a much more important one in the local SA context is that of loading.

And here the frequency used (900 or 2100) is irrelevant.

Two factors will influence your throughput based on the number of users attached to the tower.

Firstly is the concept of cell breathing, common to all 3G networks, where the size of the cell decreases rapidly as a factor of the number of users connected to it. This happens very quickly, after just a few users, and is characterised by users complaining they get good throughput during quieter times but when the network gets busy (typically at night) their performance fall away or the 3G signal completely disappears. The cell shrunk past them and they are now outside the cell.
The only way to address this is by adding more cells to ensure everyone is close to a tower.
It might appear at first that 900MHz might help here but you can see the more loaded a bigger cell becomes, the quicker it shrinks as the circumference of a larger cell will decrease more rapidly than a smaller cell.
900MHz is therefore perfect for lightly loaded rural areas but not so much urban or peri-urban environments with larger number of power users.
Fewer, larger cells have another side-effect with 'cell-overlap' much more significant than with lots of smaller cells. This overlap causes interference and other problems, again resulting in poor performance. And, again, the only answer is more but smaller cells. Operators will often 'tilt' the antennas down to make a cell smaller. This will result in gaps and users with no coverage so you then fill in the gaps with more towers.
Thus the tendency in urban and peri-urban is for more, smaller cells rather than fewer, larger ones. This is not just advantageous but critical when it comes to power users.
It's these very same power users that bring up the second problem caused by cell density, or rather, the lack thereof viz. throughput.
A single 3G tower can only handle a specific throughput and this must be shared by all users attached to it. The maximum throughput is a function of the 'modulation' and not frequency.
In a lightly loaded environment such as rural, you can use a bigger cell to service a few users but as the number of users increase throughput will drop away quickly. This will be seen by complaints of speeds that used to be good but now are much less.
These two factors (cell breathing and number of users) conspire to destroy the performance of a 3G network and the only way around this is to build more towers.

Amount of available backhaul bandwidth at the tower site

Cell tower backhaul bandwidth can be provided by various means including fiber, traditional telco connections (T1, DS-3, etc.), or wireless backhaul. Carriers will supply bandwidth to the cell tower based on anticipated bandwidth usage. If they're doing their jobs well, backhaul bandwidth will not be a limiting factor, but I would guess there are real world scenarios in which it is a limiting factor. Backhaul bandwidth available at each tower site can easily reach 1 Gbps if it is warranted or necessary.

Aggregate capacity of the spectrum and protocol

This depends heavily on the protocol being deployed and the sheer volume of spectrum being leveraged for a given site. One cell site can deploy multiple channels each using 5, 10, 20, or 40 Mhz of spectrum on each channel. Additionally, the frequency can vary from 900 MHz to 1800 MHz, etc. Also, different protocols, such as Edge, 3G, 4G, and LTE all use different methods to process signals. Some of these are more efficient than others and can squeeze more bandwidth out of a given amount of spectrum. In general, the higher the frequency, the larger the channel width, the higher the number of channels in use at the site, and the more efficient the protocol, the more bandwidth will be available.
If demand warranted, theoretically it should be possible to build a cell tower with multiple Gigabits of available bandwidth. But of course this depends on a number of highly variable factors. My guess is that today's highest capacity cell towers can provide aggregate bandwidth in the low hundreds of megabits.

Physical capacity of the network gear in use

The actual gear deployed at a cell tower site is limited by the processor speeds and other networking factors such as ethernet link speeds, etc. Some older gear is limited to 10 Mbps ethernet links to the radio devices, while newer gear is equipped with Gigabit or fiber links. Up to date gear in a properly designed cell tower should be able to support multiple Gigabits of bandwidth.

References

Comments