It’s clearly emerges that most Smart City services are based on a centralized architecture, where a dense and heterogeneous set of peripheral devices deployed over the urban area generate different types of data that are then delivered through suitable communication technologies to a control center, where data storage and processing are performed.
A primary characteristic of an urban IoT infrastructure, hence, is its capability of integrating different technologies with the existing communication infrastructures in order to support a progressive evolution of the IoT, with the interconnection of other devices and the realization of novel functionalities and services. Another fundamental aspect is the necessity to make (part of) the data collected by the urban IoT easily accessible by authorities and citizens, to increase the responsiveness of authorities to city problems, and to promote the awareness and the participation of citizens in public matters
In the rest of this section, we describe the different components of an urban IoT system, as sketched in Fig. 1. We start describing the web service approach for the design of IoT services, which requires the deployment of suitable protocol layers in the different elements of the network, as shown in the protocol stacks depicted in Fig. 1, besides the key elements of the architecture. Then, we briefly overview the link layer technologies that can be used to interconnect the different parts of the IoT. Finally, we describe the heterogeneous set of devices that concur to the realization of an urban IoT.
A. Web Service Approach for IoT Service Architecture
Although in the IoT domain many different standards are still struggling to be the reference one and the most adopted, in this section we focus specifically on IETF standards because they are open and royalty-free, are based on Internet best practices, and can count on a wide community.
Fig. 2 shows a reference protocol architecture for the urban IoT system that entails both an unconstrained and a constrained protocol stack. The first consists of the protocols that are currently the de-facto standards for Internet communications, and are commonly used by regular Internet hosts, such as XML, HTTP, and IPv4. These protocols are mirrored in the constrained protocol stack by their low-complexity counterparts, i.e., the Efficient XML Interchange (EXI), the Constrained Application Protocol (CoAP), and 6LoWPAN, which are suitable even for very constrained devices. The transcoding operations between the protocols in the left and right stacks in Fig. 2 can be performed in a standard and low complexity manner, thus guaranteeing easy access and interoperability of the IoT nodes with the Internet. It may be worth remarking that systems that do not adopt the EXI/CoAP/6LoWPAN protocol stack can still be seamlessly included in the urban IoT system, provided that they are capable of interfacing with all the layers of the left-hand side of the protocol architecture in Fig. 2.
B. Link Layer Technologies
An urban IoT system, due to its inherently large deployment area, requires a set of link layer technologies that can easily cover a wide geographical area and, at the same time, support a possibly large amount of traffic resulting from the aggregation of an extremely high number of smaller data flows. For these reasons, link layer technologies enabling the realization of an urban IoT system are classified into unconstrained and constrained technologies. The first group includes all the traditional LAN, MAN, and WAN communication technologies, such as Ethernet, WiFi, fiber optic, broadband Power Line Communication (PLC), and cellular technologies such as UMTS and LTE. They are generally characterized by high reliability, low latency, and high transfer rates (order of Mbit/s or higher), and due to their inherent complexity and energy consumption are generally not suitable for peripheral IoT nodes.
The constrained physical and link layer technologies are, instead, generally characterized by low energy consumption and relatively low transfer rates, typically smaller than 1 Mbit/s. The more prominent solutions in this category are IEEE 802.15.4 ,  Bluetooth and Bluetooth Low Energy,8 IEEE 802.11 Low Power, PLC , NFC and RFID . These links usually exhibit long latencies, mainly due to two factors: 1) the intrinsically low transmission rate at the physical layer and 2) the power-saving policies implemented by the nodes to save energy, which usually involve duty cycling with short active periods.
We finally describe the devices that are essential to realize an urban IoT, classified based on the position they occupy in the communication flow.
1) Backend Servers
At the root of the system, we find the backend servers, located in the control center, where data are collected, stored, and processed to produce added-value services. In principle, backend servers are not mandatory for an IoT system to properly operate, though they become a fundamental component of an urban IoT where they can facilitate the access to the smart city services and open data through the legacy network infrastructure. Backend systems commonly considered for interfacing with the IoT data feeders include the following.
Moving toward the “edge” of the IoT, we find the gateways, whose role is to interconnect the end devices to the main communication infrastructure of the system. With reference to the conceptual protocol architecture depicted in Fig. 2, the gateway is hence required to provide protocol translation and functional mapping between the unconstrained protocols and their constrained counterparts, that is to say XML-EXI, HTTP-CoAP, IPv4/v6-6LoWPAN
3) IoT Peripheral Nodes
Finally, at the periphery of the IoT system, we find the devices in charge of producing the data to be delivered to the control center, which are usually called IoT peripheral nodes or, more simply, IoT nodes. Generally speaking, the cost of these devices is very low, starting from 10 USD or even less, depending on the kind and number of sensors/actuators mounted on the board. IoT nodes may be classified based on a wide number of characteristics, such as powering mode, networking role (relay or leaf), sensor/actuator equipment, and supported link layer technologies. The most constrained IoT nodes are likely the Radio Frequency tags (RFtags) that, despite their very limited capabilities, can still play an important role in IoT systems, mainly because of the extremely low cost and the passive nature of their communication hardware, which does not require any internal energy source. The typical application of RFtags is object identification by proximity reading, which can be used for logistics, maintenance, monitoring, and other services.