US Region

Grandmetric LLC
Lewes DE 19958
16192 Coastal Hwy USA
EIN: 98-1615498
+1 302 691 94 10

EMEA Region

ul. Metalowa 5, 60-118 Poznań, Poland
NIP 7792433527
+48 61 271 04 43


Grandmetric LTD
Office 584b
182-184 High Street North
E6 2JA
+44 20 3321 5276

  • en
  • pl
  • Mobile network architecture – 4G design issues

    Mobile network architecture – 4G design issues

    Date: 24.09.2015

    Category: Mobile Networks

    This time I’d like to sketch the current LTE mobile network architecture. After the basic introduction of nodes and interfaces in-between them I’ll underline the most important aspects in terms of network design. The design issues are discussed in the context of radio interface and the IP-based interfaces within RAN and CN.


    LTE network architecture

    The figure below shows a sketch of the simplified “LTE” architecture (LTE really corresponds to radio-network). The whole architecture, that is shown in this figure, is an EPS (Evolved Packet System) consisting of: E-UTRAN (Evolved Universal Terrestrial Radio Access Network) and EPC (Evolved Packet Core).

    RAN consists of eNBs (Evolved NodeB), a base stations that have both, the radio interface towards terminals, and IP-based interfaces towards other base stations and towards CN (core network).

    CN consists of: nodes for user plane – including SGWs (Serving Gateway) and PGWs (Packet Data Network Gateway); and nodes responsible for control plane handling – including MME (Mobility Management Entity) and HSS (Home Subscriber Server).

    The main functions of each of the nodes are:

    • eNB – radio connectivity assurance (radio resource management, data processing, RAN security)
    • SGW – IP traffic routing between eNBs and PGW (QoS assurance, IP mobility anchor)
    • PDN GW (PGW) – IP access termination within EPC (IP address management, APN management, EPC policy enforcement)
    • MME – EPC connectivity assurance (mobility, EPC security, temporary UE management)
    • HSS – subscriber database (security and subscription profiles)


    Mobile network architecture - 4G design issues

    The interfaces that are shown in-between nodes are all IP-based. What that means, is that they are logical interfaces, which don’t necessary relate to the point-to-point connectivity.

    Traffic types related to each interface are:

    • X2 – X2-UP: IP service traffic during handoff, X2-CP: handoff based signalling, SON related signalling;
    • S1 – S1-UP (towards SGW): IP service traffic; S1-CP (towards MME): mobility / handoff based signalling, security related signalling, IP service connection management signalling, SON related signalling;
    • S11 – IP service connection management signalling;
    • S5 – IP service traffic, policy and IP service connection management signalling;
    • S6a – subscriber and APN-related data.

    Below in the next two sections we extract some important points related to the design of the described radio network and core network, that I believe, are crucial in “4G” network design.


    Design issues for radio interface / RAN

    The typical radio network design is performed within RNP (radio network planning) steps, i.e. selecting proper number of nodes (base stations), their locations and parameters (e.g. power, antenna tilt, antenna azimuth, control channel configuration, etc.) to assure required capacity and coverage. The typical approach is to have as low number of sites and equipment as possible to have reliable and profitable network (low costs of equipment, radio bandwidth and management versus high revenue from services). In recent years we observe a shift from “good old” RNP manual methods towards RAN self-organization (i.e. the use of SON). This is especially important for HetNet, where MNO (mobile network operator) need to take into account, that there are different types of nodes, where some of them (and a lot of them in the same time) are not managed directly by MNO (e.g. Home eNBs), thus making e.g. configuration or interference management more difficult or even impossible. Among others SON features “promise” the MNO to: decrease expenses on equipment (CAPEX – capital expenditures), by having optimal NW operation and thus capital investment postpone on new nodes; decrease expenses on NW operations and management (OPEX – operational expenditures), by reacting on NW behavior automatically and very fast, thus no needing to react manually be MNO’s personnel. In the area of network automation, the typical design concerns include:

    • SON features themselves (scope, configuration, GUIs, applicability, vendor independence, etc);
    • SON features interoperability (cooperation with other SON functions of the same or different SON vendor);
    • SON coordination (conflicts resolving and proper distribution of parameters changes);
    • SON features introduction (personnel trainings, from manual-SON to fully-automatic-SON gradual introduction possibility);
    • MNOs convincing to SON (OAM and RNP paradigms and procedures shift towards automation).

    Another area of the radio interface design include the management of services, that the network is to provide. Just one example is MTC (machine type communications). In recent years we observe, that MTC services are gaining more and more attention. Some of these services has a very different behavior than the legacy ones, e.g. low data rate is required and rather uplink is utilized, where e.g. typical internet browsing is data rate heavy with rather downlink utilization. The main point of this example is that we have more and more services, and service differentiation needs to be taken into account while designing radio interface for proper network operation and network congestion avoidance. The main design issues include:

    • QoS-based radio configuration;
    • Service-properties incorporation within parameters configuration;
    • Signalling optimization;
    • Incorporating the view of the future features of the radio network (e.g. LTE-Advanced features, such as Carrier Aggregation or Cooperative Multipoint Tx/Rx).


    Design issues for IP-based interfaces / CN

    The EPC is defined in a flexible and open architecture with the possibility to flexibly use and configure CN nodes (MME, SGW and PGW). What that means is, that we don’t need to build a hierarchical structure, but group nodes in terms of logical structure. That enables to manage the network efficiently in terms of redundancy of the nodes, their location and capacities. We may use e.g. several MMEs – grouped in so called “MME pools” to serve a group of eNBs simultaneously. This enables network sharing, improves network reliability, enables gradual network extensions, etc. We may also e.g. leave the eNBs to create autonomously X2 connections in-between themselves (using ANR – automatic neighbour relation function).  We may create several IP connection paths between CN nodes and between RAN and CN nodes to improve reliability of the overall network. All of this is possible, since we use the concept of IP backhaul, benefiting from IP flexibility and automatic IP mechanisms (routing, self-healing). However, most implementations are being or will be done with attention to use existing environments, thus IP backhaul concept gives us not only a way to implement the logical paths and instances, also gives possibility to use existing Layer 2 and Layer 3 infrastructures. Here comes another design challenge: how to organize X2 and S1 backhaul? As the standard does not specify what technology should seat within X2 and S1 we can imagine Layer 2 as Metro Ethernet ring using VLANs as well as typical Layer 3 architecture using e.g. MPLS L3 VPN.  Both of them are technically possible and both of them have pros and cons. What we can observe, is the tendency where the decision comes directly from existing business and architectures, but what we can suggest while building such architecture from scratch is to consider all future functions and aspects like:

    • IPv4, IPv6 convergence and concurrent use;
    • IP Backhaul scalability;
    • Traffic security (security within CN, e.g. NDS mechanisms);
    • 3rd party IP network use, traffic offloading;
    • Redundancy;
    • QoS assurance (DiffServ proper settings);
    • Any to any communication;
    • Use of MAC address , use of IP address to identify the end point;
    • Ease of services provisioning;
    • Virtualization mechanisms.


    This article presents briefly an example of “4G” mobile network architecture, and presents a set of design issues for it. The main points are:

    • the radio resources are expensive and also scarce, thus need to be configured and managed efficiently;
    • IP backhaul that is also expensive and needs to be very reliable, thus also needs to be configured and managed efficiently.


    Marcin Dryjanski, Ph.D.

    Marcin Dryjanski received his Ph.D. in telecommunications from the Poznan University of Technology in September 2019. During the past 15 years, Marcin has served as R&D Engineer, Lead Researcher, R&D Consultant, Technical Trainer, Technical Leader and Board Member. He has been providing expert-level courses in the area of 5G/LTE/LTE-Advanced for leading mobile operators and vendors. In addition to that, Marcin was a work-package leader in EU-funded research projects aiming at radio interface design for 5G including FP-7 5GNOW and FP-7 SOLDER. He co-authored a number of research papers targeting 5G radio interface design and a book "From LTE to LTE-Advanced Pro and 5G" published by Artech House. Marcin is co-founder of Grandmetric and co-founder and CEO at Rimedo Labs, currently focusing on Open RAN systems.

    Beranard. V. Babu
    5 November 2023 at 19:54

    Please let us know how much it will cost one to be trained by yourself in radio planning?

    Best regards,

    Bernard V. Babu

    Marcin Bialy
    12 December 2023 at 13:13

    Hi Bernard, please write at


    Leave a Reply

    Your email address will not be published. Required fields are marked *